<?xml version='1.0' encoding='UTF-8'?><?xml-stylesheet href="http://www.blogger.com/styles/atom.css" type="text/css"?><feed xmlns='http://www.w3.org/2005/Atom' xmlns:openSearch='http://a9.com/-/spec/opensearchrss/1.0/' xmlns:georss='http://www.georss.org/georss' xmlns:gd='http://schemas.google.com/g/2005' xmlns:thr='http://purl.org/syndication/thread/1.0'><id>tag:blogger.com,1999:blog-8830511825665412342</id><updated>2011-12-29T21:51:00.576+08:00</updated><category term='Data integration'/><category term='Web application development'/><category term='recommendation'/><category term='schama'/><category term='opinion extraction'/><category term='Don'/><category term='document classification'/><category term='Schema matching'/><category term='document clustering'/><category term='query translation'/><category term='semantic web'/><category term='aboutlab'/><category term='ontology'/><category term='web modeling'/><category term='association rule'/><category term='巫孟倫'/><category term='coclustering document'/><category term='web data extraction'/><category term='Path Detection'/><category term='information extraction'/><category term='AI'/><category term='text summarization'/><category term='JhihMing'/><category term='mobile computing'/><category term='GPS'/><category term='keyword selection'/><category term='web UI'/><category term='DBN'/><category term='cfp'/><category term='social network'/><category term='google'/><title type='text'>Web Intelligence and Data Mining Laboratory</title><subtitle type='html'>&lt;a href="http://groups.google.com/group/group-meeting/topics"&gt;討論群組&lt;/a&gt; | &lt;a href="http://progressreport4all.blogspot.com/"&gt;進度報告&lt;/a&gt; | 
&lt;a href="http://web4seminar.pbwiki.com/"&gt;文件編輯&lt;/a&gt;
| &lt;a href="http://sites.google.com/site/nculab/"&gt;English Version&lt;/a&gt;</subtitle><link rel='http://schemas.google.com/g/2005#feed' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/posts/default'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default?max-results=100'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/'/><link rel='hub' href='http://pubsubhubbub.appspot.com/'/><link rel='next' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default?start-index=101&amp;max-results=100'/><author><name>Jahui</name><uri>http://www.blogger.com/profile/04407009593178832508</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><generator version='7.00' uri='http://www.blogger.com'>Blogger</generator><openSearch:totalResults>106</openSearch:totalResults><openSearch:startIndex>1</openSearch:startIndex><openSearch:itemsPerPage>100</openSearch:itemsPerPage><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-8032394681993814352</id><published>2011-12-29T21:50:00.000+08:00</published><updated>2011-12-29T21:51:00.623+08:00</updated><title type='text'>Integrating a Piecewise Linear Representation Method and a Neural Network Model for Stock Trading Points Prediction</title><summary type='text'>Integrating a piecewise linear representation method and a neural network model for stock trading points predictionView more presentations from lolokikipipi.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/8032394681993814352/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=8032394681993814352' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8032394681993814352'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8032394681993814352'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/12/integrating-piecewise-linear.html' title='Integrating a Piecewise Linear Representation Method and a Neural Network Model for Stock Trading Points Prediction'/><author><name>黃俞翔</name><uri>http://www.blogger.com/profile/10507331966998533930</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-2578927031310955722</id><published>2011-12-13T20:03:00.001+08:00</published><updated>2011-12-13T20:06:33.101+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='巫孟倫'/><title type='text'>Collaborative Filtering with CCAM</title><summary type='text'>Collaborative filtering with CCAMView more presentations from AllenWu.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/2578927031310955722/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=2578927031310955722' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2578927031310955722'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2578927031310955722'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/12/collaborative-filtering-with-ccam.html' title='Collaborative Filtering with CCAM'/><author><name>Allen</name><uri>http://www.blogger.com/profile/03818983027330935523</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='32' height='24' src='http://1.bp.blogspot.com/_PAjpaFNp8QQ/TAdiSBkfojI/AAAAAAAAAcg/EMFUcPghtJo/S220/PB250019.JPG'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-3479128478215055792</id><published>2011-12-09T08:19:00.002+08:00</published><updated>2011-12-09T08:22:01.713+08:00</updated><title type='text'>大咪</title><summary type='text'> Personalizing web page recommendation via collaborative filtering and   View more presentations from johnnyne  </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/3479128478215055792/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=3479128478215055792' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3479128478215055792'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3479128478215055792'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/12/blog-post.html' title='大咪'/><author><name>強尼</name><uri>http://www.blogger.com/profile/02782424041080171803</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-5881809429273230575</id><published>2011-12-02T09:21:00.002+08:00</published><updated>2011-12-07T16:55:04.893+08:00</updated><title type='text'>A Multi-Agent Prediction Market Based on Boolean Network Evolution</title><summary type='text'> A multi agent prediction market based on Boolean Network Evolution   View more presentations from lolokikipipi  </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/5881809429273230575/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=5881809429273230575' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5881809429273230575'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5881809429273230575'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/12/multi-agent-prediction-market-based-on.html' title='A Multi-Agent Prediction Market Based on Boolean Network Evolution'/><author><name>黃俞翔</name><uri>http://www.blogger.com/profile/10507331966998533930</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-3379553368491711622</id><published>2011-12-01T16:51:00.001+08:00</published><updated>2011-12-01T16:51:59.810+08:00</updated><title type='text'></title><summary type='text'>Mining group correlations over data streamsView more presentations from yuanchung.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/3379553368491711622/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=3379553368491711622' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3379553368491711622'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3379553368491711622'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/12/mining-group-correlations-over-data_01.html' title=''/><author><name>Jeff</name><uri>http://www.blogger.com/profile/03979400308457621747</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-8165435924660426446</id><published>2011-11-24T17:23:00.001+08:00</published><updated>2011-11-24T17:24:30.855+08:00</updated><title type='text'>Large Scale Text classification using Semi-supervised MNB</title><summary type='text'>MLView more presentations from 慶治 陳慶治.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/8165435924660426446/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=8165435924660426446' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8165435924660426446'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8165435924660426446'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/11/large-scale-text-classification-using.html' title='Large Scale Text classification using Semi-supervised MNB'/><author><name>陳大白</name><uri>http://www.blogger.com/profile/02614616959182820362</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-8123961930430398197</id><published>2011-11-17T21:17:00.000+08:00</published><updated>2011-11-17T21:18:22.854+08:00</updated><title type='text'>Query dependent ranking using k nearest neighbor</title><summary type='text'> Query dependent ranking using k nearest neighbor   View more presentations from iyo  </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/8123961930430398197/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=8123961930430398197' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8123961930430398197'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8123961930430398197'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/11/query-dependent-ranking-using-k-nearest.html' title='Query dependent ranking using k nearest neighbor'/><author><name>IYO</name><uri>http://www.blogger.com/profile/11251684987125700139</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-6708137423222743096</id><published>2011-11-03T19:12:00.002+08:00</published><updated>2011-12-13T20:07:40.882+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='巫孟倫'/><title type='text'>DSTree: A Tree Structure for the Mining of Frequent Sets from Data Streams</title><summary type='text'> DSTree: A Tree Structure for the Mining of Frequent Sets from Data Streams   View more presentations from AllenWu  </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/6708137423222743096/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=6708137423222743096' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/6708137423222743096'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/6708137423222743096'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/11/dstree-tree-structure-for-mining-of.html' title='DSTree: A Tree Structure for the Mining of Frequent Sets from Data Streams'/><author><name>Allen</name><uri>http://www.blogger.com/profile/03818983027330935523</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='32' height='24' src='http://1.bp.blogspot.com/_PAjpaFNp8QQ/TAdiSBkfojI/AAAAAAAAAcg/EMFUcPghtJo/S220/PB250019.JPG'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-7722368531938974445</id><published>2011-11-03T18:11:00.000+08:00</published><updated>2011-11-03T18:11:13.931+08:00</updated><title type='text'>Wsd as distributed constraint optimization problem</title><summary type='text'>Wsd as distributed constraint optimization problem  View more presentations from lolokikipipi </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/7722368531938974445/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=7722368531938974445' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7722368531938974445'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7722368531938974445'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/11/wsd-as-distributed-constraint.html' title='Wsd as distributed constraint optimization problem'/><author><name>空空</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='32' height='32' src='http://1.bp.blogspot.com/-YN-n-v-GmYU/TfCNjvMgCEI/AAAAAAAAAAk/pNQ0D71UgZU/s1600/49946_100000837718811_4664802_q.jpg'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-5444445084072485050</id><published>2011-10-24T21:16:00.001+08:00</published><updated>2011-10-24T21:16:58.072+08:00</updated><title type='text'>FRank: A Ranking method with Fidelity Loss</title><summary type='text'> FRank: A Ranking Method with Fidelity Loss   View more presentations from 體妮 陳  </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/5444445084072485050/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=5444445084072485050' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5444445084072485050'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5444445084072485050'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/10/frank-ranking-method-with-fidelity-loss.html' title='FRank: A Ranking method with Fidelity Loss'/><author><name>Deanli</name><uri>http://www.blogger.com/profile/01431582102464377783</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-7866189119373508256</id><published>2011-10-11T15:02:00.000+08:00</published><updated>2011-10-11T15:03:06.151+08:00</updated><title type='text'>Discovering Organizational Structure in Dynamic Social Network</title><summary type='text'> 2011 10-14 大咪報告   View more presentations from chenbojyh  </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/7866189119373508256/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=7866189119373508256' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7866189119373508256'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7866189119373508256'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/10/discovering-organizational-structure-in.html' title='Discovering Organizational Structure in Dynamic Social Network'/><author><name>Po-Chih</name><uri>http://www.blogger.com/profile/11076273327805380340</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-2180784634880521058</id><published>2011-10-07T11:07:00.001+08:00</published><updated>2011-10-07T11:09:19.250+08:00</updated><title type='text'>Mining top-k frequent closed itemsets over data streams using the sliding window model</title><summary type='text'>Mining top k frequent closed itemsetsView more presentations from yuanchung.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/2180784634880521058/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=2180784634880521058' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2180784634880521058'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2180784634880521058'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/10/mining-top-k-frequent-closed-itemsets.html' title='Mining top-k frequent closed itemsets over data streams using the sliding window model'/><author><name>Jeff</name><uri>http://www.blogger.com/profile/03979400308457621747</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-2911659435232743113</id><published>2011-09-30T08:04:00.000+08:00</published><updated>2011-09-30T08:11:02.444+08:00</updated><title type='text'>I Want to Answer, Who Has a Question? Yahoo! Answers Recommender System</title><summary type='text'>I want to answer, who has aView more presentations from chenbojyh.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/2911659435232743113/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=2911659435232743113' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2911659435232743113'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2911659435232743113'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/09/i-want-to-answer-who-has-question-yahoo.html' title='I Want to Answer, Who Has a Question? Yahoo! Answers Recommender System'/><author><name>Po-Chih</name><uri>http://www.blogger.com/profile/11076273327805380340</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-7860566247171864916</id><published>2011-09-29T18:57:00.002+08:00</published><updated>2011-09-29T19:30:32.000+08:00</updated><title type='text'>A wen usage mining approach based on lcs algorithm in online prediction recommendation system</title><summary type='text'> 大咪報告 View more presentations from johnnyne </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/7860566247171864916/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=7860566247171864916' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7860566247171864916'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7860566247171864916'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/09/wen-usage-mining-approach-based-on-lcs.html' title='A wen usage mining approach based on lcs algorithm in online prediction recommendation system'/><author><name>強尼</name><uri>http://www.blogger.com/profile/02782424041080171803</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-8499185412464135226</id><published>2011-09-19T17:19:00.001+08:00</published><updated>2011-09-19T22:49:54.806+08:00</updated><title type='text'>Amnesic Neural Network for Classification: Application on Stock Trend Prediction*</title><summary type='text'> Amnestic neural network for classification   View more presentations from lolokikipipi  </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/8499185412464135226/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=8499185412464135226' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8499185412464135226'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8499185412464135226'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/09/amnesic-neural-network-for.html' title='Amnesic Neural Network for Classification: Application on Stock Trend Prediction*'/><author><name>黃俞翔</name><uri>http://www.blogger.com/profile/10507331966998533930</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-2093230397614109583</id><published>2011-08-24T16:05:00.000+08:00</published><updated>2011-08-24T16:06:27.745+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='巫孟倫'/><title type='text'>Co-clustering with augmented data</title><summary type='text'> Co-clustering with augmented data   View more presentations from AllenWu  </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/2093230397614109583/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=2093230397614109583' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2093230397614109583'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2093230397614109583'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/08/co-clustering-with-augmented-data.html' title='Co-clustering with augmented data'/><author><name>Allen</name><uri>http://www.blogger.com/profile/03818983027330935523</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='32' height='24' src='http://1.bp.blogspot.com/_PAjpaFNp8QQ/TAdiSBkfojI/AAAAAAAAAcg/EMFUcPghtJo/S220/PB250019.JPG'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-8585809635210957936</id><published>2011-04-26T11:16:00.002+08:00</published><updated>2011-04-26T11:31:35.901+08:00</updated><title type='text'>Ch4.mapreduce algorithm design</title><summary type='text'>Chapter 4 of Data-Intensive Text Processing with Map Reduce introduce the efficiently algorithms, pairs and stripes. It display how to use these algorithms to construct the co-occurrence matrix and how to use this matrix to compute the conditional probability. They compare the time complexity between pairs and stripes algorithms. The stripes algorithms can achieve the better efficiency than pairs</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/8585809635210957936/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=8585809635210957936' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8585809635210957936'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8585809635210957936'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/04/ch4mapreduce-algorithm-design.html' title='Ch4.mapreduce algorithm design'/><author><name>Allen</name><uri>http://www.blogger.com/profile/03818983027330935523</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='32' height='24' src='http://1.bp.blogspot.com/_PAjpaFNp8QQ/TAdiSBkfojI/AAAAAAAAAcg/EMFUcPghtJo/S220/PB250019.JPG'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-2632356397073514106</id><published>2011-01-25T17:03:00.001+08:00</published><updated>2011-01-25T17:26:26.813+08:00</updated><title type='text'>AUTOMATIC CONTENT TARGETING ON MOBILE PHONES</title><summary type='text'>因電信市場已日趨飽和，例如歐洲使用手機人口幾乎達到100%，所以電信業者將競爭力轉向加值服務(VAS)，期望能有更多收入。但是要如何有效管理數量龐大的VAS，才能大幅提高收入呢？此篇論文提出一個成功的自動化系統，能將最相關的VAS優惠訊息發送給潛在的使用者，藉此幫助電信業者實際增加收益。作者遇到之問題值得我們參考，諸如處理大量的VAS優惠訊息、和每位客戶接觸的機會有限(每天只發送一封MMS，內容包含1至4個VAS優惠訊息)、設備限制(不能達到完全一對一客製化發送訊息，需要將使用者分群)、VAS分類問題等等。此系統最核心之處為根據使用者過去購買紀錄，使用Spherical k-means演算法作使用者分群，並加入最佳化方法。值得注意的是使用者會隨著時間而有不同的興趣，作者實際測試結果，若一直給予使用者同一種廣告，使用者會對此失去興趣，並反映在實驗數據上，所以提供多樣VAS優惠訊息，</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/2632356397073514106/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=2632356397073514106' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2632356397073514106'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2632356397073514106'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/01/automatic-content-targeting-on-mobile.html' title='AUTOMATIC CONTENT TARGETING ON MOBILE PHONES'/><author><name>浪漫痕跡</name><uri>http://www.blogger.com/profile/15058398033679964366</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-7793287155418791715</id><published>2011-01-14T16:33:00.002+08:00</published><updated>2011-01-14T16:33:29.350+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='JhihMing'/><title type='text'>Adaptive web page content identification</title><summary type='text'>Adaptive web page content identification
View more presentations from JhihMing.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/7793287155418791715/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=7793287155418791715' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7793287155418791715'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7793287155418791715'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2011/01/adaptive-web-page-content.html' title='Adaptive web page content identification'/><author><name>JhihMing</name><uri>http://www.blogger.com/profile/00801853076135212256</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-5950001123623928080</id><published>2010-11-26T10:51:00.002+08:00</published><updated>2010-11-26T10:53:36.357+08:00</updated><title type='text'>最新消息</title><summary type='text'></summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/5950001123623928080/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=5950001123623928080' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5950001123623928080'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5950001123623928080'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/11/blog-post.html' title='最新消息'/><author><name>Jahui</name><uri>http://www.blogger.com/profile/04407009593178832508</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-68202974737041098</id><published>2010-11-09T13:45:00.002+08:00</published><updated>2010-11-11T23:29:25.811+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='巫孟倫'/><title type='text'>Collaborative filtering using orthogonal nonnegative matrix</title><summary type='text'>Check out this SlideShare Presentation: Collaborative filtering using orthogonal nonnegative matrixView more presentations from AllenWu.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/68202974737041098/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=68202974737041098' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/68202974737041098'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/68202974737041098'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/11/collaborative-filtering-using.html' title='Collaborative filtering using orthogonal nonnegative matrix'/><author><name>Allen</name><uri>http://www.blogger.com/profile/03818983027330935523</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='32' height='24' src='http://1.bp.blogspot.com/_PAjpaFNp8QQ/TAdiSBkfojI/AAAAAAAAAcg/EMFUcPghtJo/S220/PB250019.JPG'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-5425661473519750855</id><published>2010-10-26T14:21:00.000+08:00</published><updated>2010-10-26T14:22:01.906+08:00</updated><title type='text'>A Scalable Collaborative Filtering Framework based on Co-clustering</title><summary type='text'>A scalable collaborative filtering framework based on co-clusteringView more presentations from lau.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/5425661473519750855/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=5425661473519750855' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5425661473519750855'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5425661473519750855'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/10/scalable-collaborative-filtering.html' title='A Scalable Collaborative Filtering Framework based on Co-clustering'/><author><name>Rui</name><uri>http://www.blogger.com/profile/00071307474432848848</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-1138555698687483006</id><published>2010-10-12T23:24:00.002+08:00</published><updated>2010-10-12T23:26:32.436+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='JhihMing'/><title type='text'>Content extraction via tag ratios</title><summary type='text'>Content extraction via tag ratios
View more presentations from JhihMing.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/1138555698687483006/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=1138555698687483006' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1138555698687483006'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1138555698687483006'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/10/content-extraction-via-tag-ratios.html' title='Content extraction via tag ratios'/><author><name>JhihMing</name><uri>http://www.blogger.com/profile/00801853076135212256</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-5082194738624968434</id><published>2010-09-24T00:45:00.006+08:00</published><updated>2010-10-12T23:28:25.714+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='JhihMing'/><title type='text'>Extracting article text from the web with maximum subsequence segmentation</title><summary type='text'>



Extracting article text from the web with maximum subsequence segmentation
View more presentations from JhihMing.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/5082194738624968434/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=5082194738624968434' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5082194738624968434'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5082194738624968434'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/09/extracting-article-text-from-web-with.html' title='Extracting article text from the web with maximum subsequence segmentation'/><author><name>JhihMing</name><uri>http://www.blogger.com/profile/00801853076135212256</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-7540502225560527306</id><published>2010-09-11T14:57:00.000+08:00</published><updated>2010-09-11T14:58:06.904+08:00</updated><title type='text'>Bayesian Co-clustering</title><summary type='text'>Presented by 睿哲:Bayesian Co clusteringView more presentations from lau.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/7540502225560527306/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=7540502225560527306' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7540502225560527306'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7540502225560527306'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/09/bayesian-co-clustering.html' title='Bayesian Co-clustering'/><author><name>Rui</name><uri>http://www.blogger.com/profile/00071307474432848848</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-5629775186272409743</id><published>2010-09-07T19:49:00.001+08:00</published><updated>2010-09-07T19:57:47.158+08:00</updated><title type='text'>Does BT truly have the ability to help online advertising?</title><summary type='text'>網路廣告最常使用之技術是Behavioral Targeting (BT)，且BT在網路廣告領域扮演越來越重要的角色。但學術界尚未證明BT是否真的對網路廣告有多大幫助。本篇論文之目標是證明BT對網路廣告有多少幫助，並將此問題分成三個子問題循序論證：(1)BT是否能幫助網路廣告?(2)利用evaluation metrics，BT可幫助網路網告多少?(3)何種BT策略較佳?BT之假設為若使用者具相似搜尋或瀏覽行為，則使用者點選相同廣告之機率較高。如此一來，針對第一個問題，作者先將使用者分成within-及between- ads user similarity兩群，之後用Cosine similarity計算彼此之相似度，發現點選相同廣告使用者之相似度為點選不同廣告使用者相似度90倍，故可證明BT真的對網路廣告有幫助。藉由計算將使用者分成K群前之Click-Through Rate (</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/5629775186272409743/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=5629775186272409743' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5629775186272409743'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5629775186272409743'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/09/does-bt-truly-have-ability-to-help.html' title='Does BT truly have the ability to help online advertising?'/><author><name>浪漫痕跡</name><uri>http://www.blogger.com/profile/15058398033679964366</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-6612262630147478159</id><published>2010-08-17T16:34:00.006+08:00</published><updated>2010-08-17T22:29:21.312+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='schama'/><title type='text'>Survey of Schema Mapping Approach</title><summary type='text'>此篇論文為綱要對映領域的Survey性質的論文，依據資料來源等特性對於現有的綱要對映做分類。　　綱要對映是許多研究資訊整合所必須面臨的重要問題，舉凡電子商務、資料倉儲、語意查詢等均需要綱要對映來合併不同來源的資料。此論文就所參考的資料來源作綱要層級和實例層級的分類，綱要層級表示所參考的資料均使用綱要所包含的資訊（綱要名稱、綱要描述、綱要限制等），並不包括實例層級的資訊；而實例層級則只考慮真實世界中的例子和其屬性來當作綱要對映的依據，通常用以修正綱要層級所無法察覺的資訊或者輔助其準確率。當然，實例層級本身也可用於綱要對映。接著再以對映的元素大小作為元素層級和架構層級的分類，元素層級表示每次對映時都以單一元素和單一元素作對映；架構層級則必須考慮相同記錄中元素間相互連結的關係，使之與另一記錄的架構作對映。細分下去就分為語意層級和限制層級，語意層級使用元素的文字內容相似度當作對映的依據，</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/6612262630147478159/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=6612262630147478159' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/6612262630147478159'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/6612262630147478159'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/08/survey-of-schema-mapping-approach.html' title='Survey of Schema Mapping Approach'/><author><name>Kuan-Chi-I</name><uri>http://www.blogger.com/profile/00305559112409878282</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><media:thumbnail xmlns:media='http://search.yahoo.com/mrss/' url='http://2.bp.blogspot.com/_yylvKtzmFmk/TGpJxYAkB5I/AAAAAAAAABY/ocSa_PPo1kE/s72-c/0.bmp' height='72' width='72'/><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-7223327286900926802</id><published>2010-08-12T12:20:00.004+08:00</published><updated>2010-08-17T21:46:34.534+08:00</updated><title type='text'>Address Standardization with Latent Semantic Association</title><summary type='text'>        地址的表示方式通常並不具有規則性，因為它通常是由不同人在不同時期所寫的。所以地址標準化就是具有高重要性的工作。監督方式是需要大量的已標記訓練資料，而對每個目標領域建設一個大規模標註語料庫是一個費時且費工的任務，所以為了最大限度地減少人的為的影響與標記訓練資料集的大小，本篇論文提出了一個地址標準化法與潛在語義聯想（LaSA）之模型。        LaSA模型是建構來抓取未標記語料集中潛在語義的聯想字詞。首先，目標區域中的原始字詞空間會被投影到使用LaSA模型的概念空間，然後地址標準化模型會主動地從LaSA模型特徵和豐富的訓練資料中被學習出來，而方法能有效且正確地抓取文件中的資料。針對以大型英語和漢語語料庫為對象的實驗結果顯示，此方法提高了標準化工作的性能與可以僅使用較少的訓練資料就達到較好效能之功效。Address standardization with latent </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/7223327286900926802/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=7223327286900926802' title='2 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7223327286900926802'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7223327286900926802'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/08/address-standardization-with-latent.html' title='Address Standardization with Latent Semantic Association'/><author><name>JiaYi</name><uri>http://www.blogger.com/profile/07667017703017273245</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>2</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-9079339564688583019</id><published>2010-07-27T16:37:00.000+08:00</published><updated>2010-07-27T16:38:15.246+08:00</updated><title type='text'>Providing location based information/advertising for existing mobile phone users</title><summary type='text'>本篇論文提出根據位置提供相關訊息／廣告至行動電話之系統，只要具有藍芽功能之行動電話，且不需安裝軟體，即可收到與位置相關的訊息／廣告。此系統可應用於超市廣告／優惠卷系統(Supermarket adverts/coupons)及導遊系統(Guide system)。近年使用藍芽的人數越來越多，根據諾基亞(Nokia)調查2006年使用人數增加了65%，可見愈來愈多人可接受使用藍芽傳輸接收訊息、檔案、甚至廣告。超市廣告／優惠券系統即在超市中設置若干台藍芽發送器發送超市商品訊息、廣告或優惠券到擁有藍芽功能的裝置或PDA的使用者，而且因系統不需安裝任何軟體，所以此系統可以服務多種具有藍芽功能的裝置，不因裝置種類而受限。但有個值得注意的問題是同一位使用者可能會出現在同一台藍芽發送器的附近不只一次，要避免讓相同使用者短時間內收到同樣訊息、廣告。本篇論文使用後端訊息系統(backend </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/9079339564688583019/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=9079339564688583019' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/9079339564688583019'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/9079339564688583019'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/07/providing-location-based.html' title='Providing location based information/advertising for existing mobile phone users'/><author><name>浪漫痕跡</name><uri>http://www.blogger.com/profile/15058398033679964366</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-2693876003806419395</id><published>2010-07-16T14:13:00.006+08:00</published><updated>2010-10-12T23:19:22.778+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='JhihMing'/><title type='text'>Comments oriented blog summarization by sentence extraction</title><summary type='text'>這篇paper是focus在comments-oriented blog post summarization，利用comment的information去計算word representativeness score，再基於每一個word的score去select sentence。Comments oriented blog summarization by sentence extraction
View more presentations from JhihMing.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/2693876003806419395/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=2693876003806419395' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2693876003806419395'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2693876003806419395'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/07/comments-oriented-blog-summarization-by.html' title='Comments oriented blog summarization by sentence extraction'/><author><name>JhihMing</name><uri>http://www.blogger.com/profile/00801853076135212256</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-8605011210006312628</id><published>2010-06-02T15:22:00.000+08:00</published><updated>2010-06-02T15:23:55.683+08:00</updated><title type='text'>Grouping Search-Engine Returned Citations for Person-Name Queries</title><summary type='text'>此篇論文提出了一個解決搜尋引擎回傳的搜尋結果中同名同姓的問題。用搜尋引擎查詢人名時，搜尋引擎回傳的搜尋結果常包含多位同名同姓的人，不僅沒有分門別類，還將同一個人的資訊分散在搜尋結果中，讓查詢者非常不方便。此篇論文提出Multi-Faceted Approach，對Attribute、Links及Page Similarity分別都作Confidence Matrix，再使用Stanford certainty theory將三個Confidence Matrix合併。之後，利用Final Confidence Matrix作Grouping，即可以得到同一個人的搜尋結果會在同一群組中，而且仍保持搜尋結果的資訊及編排的順序。實驗結果顯示效果頗佳。Grouping search engine returned citations for person-name queriesView </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/8605011210006312628/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=8605011210006312628' title='1 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8605011210006312628'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8605011210006312628'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/06/grouping-search-engine-returned.html' title='Grouping Search-Engine Returned Citations for Person-Name Queries'/><author><name>浪漫痕跡</name><uri>http://www.blogger.com/profile/15058398033679964366</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>1</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-4040452716786808360</id><published>2010-06-02T15:18:00.000+08:00</published><updated>2010-06-02T15:19:31.328+08:00</updated><title type='text'>Two Birds with One Stone: A Graph-based Framework for Disambiguating and Tagging People Names in Web Search</title><summary type='text'>此篇問題想解決對搜尋引擎作人名搜尋所回傳搜尋結果中同名同姓的問題。作者將屬性，包含地點、電子郵件、工作單位…等等的資訊，以圖形表示，再給予圖中的點、邊不同的權重值。然後使用分群演算法將圖做分群，即可得到每一個人正確的相關資訊。因為篇幅限制，對於實驗過程有些問題尚未了解清楚。總之，這個方法不但處理同名同姓的問題(Disambiguating)，也能將正確的人名與她正確的資訊配對(Tagging People Names)，是一石二鳥的好方法。NamesakeView more presentations from avelinhuo.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/4040452716786808360/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=4040452716786808360' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/4040452716786808360'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/4040452716786808360'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/06/two-birds-with-one-stone-graph-based.html' title='Two Birds with One Stone: A Graph-based Framework for Disambiguating and Tagging People Names in Web Search'/><author><name>浪漫痕跡</name><uri>http://www.blogger.com/profile/15058398033679964366</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-7511624822867729017</id><published>2010-05-27T00:56:00.002+08:00</published><updated>2010-06-11T05:57:07.097+08:00</updated><title type='text'>Cross-Domain Sentiment Classification via Spectral Feature Alignment</title><summary type='text'>Cross domain sentiment classification via spectral feature alignmentView more presentations from lau.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/7511624822867729017/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=7511624822867729017' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7511624822867729017'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7511624822867729017'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/05/2010cross-domain-sentiment.html' title='Cross-Domain Sentiment Classification via Spectral Feature Alignment'/><author><name>Rui</name><uri>http://www.blogger.com/profile/00071307474432848848</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-1306276095298078323</id><published>2010-05-02T19:16:00.000+08:00</published><updated>2010-06-11T05:56:08.835+08:00</updated><title type='text'></title><summary type='text'>An experimental comparison of naive bayesian and keyword basedView more presentations from eraser60913.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/1306276095298078323/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=1306276095298078323' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1306276095298078323'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1306276095298078323'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/06/experimental-comparison-of-naive.html' title=''/><author><name>林書彥</name><uri>http://www.blogger.com/profile/16694639084485641957</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-8504137685863232365</id><published>2010-04-27T10:38:00.001+08:00</published><updated>2010-06-11T05:57:27.099+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='coclustering document'/><title type='text'>Coclustering based classfication for out-of-domain documents</title><summary type='text'>Coclustering Base Classification For Out Of Domain DocumentsView more presentations from lau.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/8504137685863232365/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=8504137685863232365' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8504137685863232365'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8504137685863232365'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/04/2007coclustering-based-classfication.html' title='Coclustering based classfication for out-of-domain documents'/><author><name>Rui</name><uri>http://www.blogger.com/profile/00071307474432848848</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-3019567624130006997</id><published>2010-04-10T15:10:00.002+08:00</published><updated>2010-10-12T23:20:30.524+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='JhihMing'/><title type='text'>GHOST: An Effective Graph-based Framework for Name Distinction</title><summary type='text'>Check out this SlideShare Presentation: 
Ghost
View more presentations from JhihMing.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/3019567624130006997/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=3019567624130006997' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3019567624130006997'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3019567624130006997'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/04/ghost-effective-graph-based-framework.html' title='GHOST: An Effective Graph-based Framework for Name Distinction'/><author><name>JhihMing</name><uri>http://www.blogger.com/profile/00801853076135212256</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-8437532989553482646</id><published>2010-03-19T12:05:00.002+08:00</published><updated>2010-03-19T12:14:18.614+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='Path Detection'/><category scheme='http://www.blogger.com/atom/ns#' term='mobile computing'/><title type='text'>Learning transportation mode from raw gps data for geographic applications on the web</title><summary type='text'>本論文摘自WWW 2008, 主要貢獻在於如何從原始的GPS資料推論出當時的交通工具 (e.g., Car, Bus, Bike, Walk).方法可分為以下三個步驟: 1. change point-based segmentation method 負責將GPS的trip分割成不同的segmentation, 這裡僅區分成Walk Segment 以及 Non Walk Segment.2. inference model: 本論文比較四種不同的機器學習方法 (Navie Bayes, Decision Tree, SVM, CRT), 針對每個segment辨識所使用的交通工具. 3. post-processing algorithm based on conditional probability: 有鑒於機器學習推論出來的交通工具有所誤差, </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/8437532989553482646/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=8437532989553482646' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8437532989553482646'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8437532989553482646'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/03/learning-transportation-mode-from-raw.html' title='Learning transportation mode from raw gps data for geographic applications on the web'/><author><name>Stanley Fan 范登凱</name><uri>http://www.blogger.com/profile/00552079070299670477</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-8807452037104331592</id><published>2010-03-12T16:18:00.002+08:00</published><updated>2010-05-04T17:56:02.061+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='巫孟倫'/><title type='text'>A Generalized Maximum Entropy Approach To Bregman Co Clustering</title><summary type='text'>Check out this SlideShare Presentation: A Generalized Maximum Entropy Approach To Bregman Co ClusteringView more presentations from guest00a636.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/8807452037104331592/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=8807452037104331592' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8807452037104331592'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8807452037104331592'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/03/generalized-maximum-entropy-approach-to.html' title='A Generalized Maximum Entropy Approach To Bregman Co Clustering'/><author><name>Allen</name><uri>http://www.blogger.com/profile/03818983027330935523</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='32' height='24' src='http://1.bp.blogspot.com/_PAjpaFNp8QQ/TAdiSBkfojI/AAAAAAAAAcg/EMFUcPghtJo/S220/PB250019.JPG'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-6361383123374178927</id><published>2010-03-09T20:05:00.001+08:00</published><updated>2010-03-09T20:10:06.707+08:00</updated><title type='text'>Retrieving Address Based Locations From The Web</title><summary type='text'>當前主要的搜尋引擎處理基於關鍵字的查詢非常有效。使用者已經在使用這些功能的空間搜尋已知地名。不過，這檢索過程往往很多與空間無關的文件。一個空間網路搜尋不僅了解含有使用者查詢的地點，但更重要的是了解在網頁上的位置與使用者的查詢如何能正確地匹配。對於location-based Web search，這篇論文實施了自己的地理資訊網路搜尋引擎。其中我們採用的方法所注重抓取的是去檢索與位置有高度相關的網頁。為此，我們使用基於geoparser的基礎上的對位置做分類，以引導Clawler。Retrieving Address Based Locations From The WebView more presentations from jyhuangtc.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/6361383123374178927/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=6361383123374178927' title='1 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/6361383123374178927'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/6361383123374178927'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2010/03/retrieving-address-based-locations-from.html' title='Retrieving Address Based Locations From The Web'/><author><name>JiaYi</name><uri>http://www.blogger.com/profile/07667017703017273245</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>1</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-1959217309763565863</id><published>2009-11-16T15:53:00.002+08:00</published><updated>2009-11-16T16:03:59.548+08:00</updated><title type='text'>Conditional Random Fields : Probabilistic Models for Segmenting and Labeling Sequence Data</title><summary type='text'>主要介紹CRF的概念，參考一下兩篇相關文獻 : [1] J. Lafferty, A. McCallum, and F. Pereira. Conditional random fields: probabilistic models for segmenting and labeling sequence data. In International Conference on Machine Learning, 2001. [2] Hanna M. Wallach. Conditional Random Fields: An Introduction. University of Pennsylvania CIS Technical Report MS-CIS-04-21.過去針對sequential data進行segmenting或是labeling的動作，有HMM、MEMMs</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/1959217309763565863/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=1959217309763565863' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1959217309763565863'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1959217309763565863'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2009/11/conditional-random-fields-probabilistic.html' title='Conditional Random Fields : Probabilistic Models for Segmenting and Labeling Sequence Data'/><author><name>swing</name><uri>http://www.blogger.com/profile/10386401967984346262</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-5247719034755988052</id><published>2009-11-04T13:26:00.002+08:00</published><updated>2010-05-04T17:57:00.906+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='巫孟倫'/><title type='text'>Semantics In Digital Photos A Contenxtual Analysis</title><summary type='text'>Interpreting the semantics of an image is a hard problem.However, for storing and indexing large multimedia lections,it is essential to build systems that can automatically extract semantics from images. In this research we show how we can fuse content and context to extract semantics from digital photographs. Our experiments show that if we can properly model context associated with media, we </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/5247719034755988052/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=5247719034755988052' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5247719034755988052'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5247719034755988052'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2009/11/semantics-in-digital-photos-contenxtual.html' title='Semantics In Digital Photos A Contenxtual Analysis'/><author><name>Allen</name><uri>http://www.blogger.com/profile/03818983027330935523</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='32' height='24' src='http://1.bp.blogspot.com/_PAjpaFNp8QQ/TAdiSBkfojI/AAAAAAAAAcg/EMFUcPghtJo/S220/PB250019.JPG'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-4588908460315707575</id><published>2009-10-12T22:20:00.002+08:00</published><updated>2009-10-13T01:16:28.102+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='document clustering'/><category scheme='http://www.blogger.com/atom/ns#' term='keyword selection'/><category scheme='http://www.blogger.com/atom/ns#' term='recommendation'/><title type='text'>Matching Task Profiles and User Needs in Personalized Web Search</title><summary type='text'>本篇文章(Matching Task Profiles and User Needs in Personalized Web Search)出自於CIKM'08,本篇研究提出一套個人化網路搜尋架構, 其主要係透過以統計為基礎的語言模型方法,用它來偵測使用者在網路搜尋時的需求與意圖.此個人化架構基於使用者個人過往的歷史搜尋紀錄, 並藉此資訊來媒合使用者當前的搜尋需求, 其主要方法如下:首先, 個人化搜尋架構會將使用者目前的搜尋結果和歷史搜尋記錄進行相似度計算, 計算方法採用Kullback-Leibler (KL) divergence. 接著, 系統會利用 threshold 機制來決定是否要將搜尋查詢實施 query rewritting 或將搜尋結果進行 re-ranking 的動作.有鑒於, 相似度的計算所採用的是KL divergence方法, 因此主要係比較歷史紀錄(tasks</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/4588908460315707575/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=4588908460315707575' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/4588908460315707575'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/4588908460315707575'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2009/10/matching-task-profiles-and-user-needs.html' title='Matching Task Profiles and User Needs in Personalized Web Search'/><author><name>Stanley Fan 范登凱</name><uri>http://www.blogger.com/profile/00552079070299670477</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-2989042465817908957</id><published>2009-10-05T15:00:00.000+08:00</published><updated>2009-10-05T15:04:21.691+08:00</updated><title type='text'>Designing a Syntax-Based Retrieval System for Supporting Language Learning</title><summary type='text'>Designing A Syntax Based Retrieval System03View more presentations from guestbbe441a.這篇論文主要想處理的問題是能夠解學英文學習者在快速精確搜尋語句與文法。使用regular expression search engine 提供快速、可靠度高而且搜尋結果不錯的功能。但是，如果沒有好的index scheme 設計的話，搜尋執行的時間卻是它的嚴重致命傷。Cho and Rajagopalan (2002) 提出來的presuf index constructing algorithm 和 deciding minimum filter factor (minff) 等等，是一個不錯的方法解決搜尋時間過長的問題。他把index size 縮小，而且效能不錯。另外，本論文的作者還設計的一個</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/2989042465817908957/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=2989042465817908957' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2989042465817908957'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2989042465817908957'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2009/10/designing-syntax-based-retrieval-system.html' title='Designing a Syntax-Based Retrieval System for Supporting Language Learning'/><author><name>浪漫痕跡</name><uri>http://www.blogger.com/profile/15058398033679964366</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-3634724989437633381</id><published>2009-08-25T13:57:00.003+08:00</published><updated>2010-05-04T17:57:59.336+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='巫孟倫'/><title type='text'>Decision Support And Profit Prediction For Online Auction</title><summary type='text'>Decision Support And Profit Prediction For Online AuctionView more presentations from AllenWu.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/3634724989437633381/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=3634724989437633381' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3634724989437633381'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3634724989437633381'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2009/08/decision-support-and-profit-prediction.html' title='Decision Support And Profit Prediction For Online Auction'/><author><name>Allen</name><uri>http://www.blogger.com/profile/03818983027330935523</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='32' height='24' src='http://1.bp.blogspot.com/_PAjpaFNp8QQ/TAdiSBkfojI/AAAAAAAAAcg/EMFUcPghtJo/S220/PB250019.JPG'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-6707712430521415196</id><published>2009-08-23T21:17:00.002+08:00</published><updated>2009-08-23T21:21:43.675+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='keyword selection'/><category scheme='http://www.blogger.com/atom/ns#' term='document classification'/><title type='text'>Detecting Online Commercial Intention (OCI)</title><summary type='text'>本篇文章 (Detecting Online Commercial Intention (OCI)) 出自於WWW'06, 主要的作者為微軟研究團隊. 如果可以瞭解或預測使用者的網路行為, 則可以進一步給予個人化內容, 例如: 呈現使用者喜愛的產品或新聞內容給使用者. 因此對資訊服務提供者 (Information Service Provider) 而言能夠精確的預測使用者行為, 係一項很重要的任務. 本論文主要探討使用者在網路上面的行為是否具有商業意圖, 稱為 Online Commercial Intention (OCI), 作者將網路行為主要分成使用者正在瀏覽的網頁 (web) 和使用者所搜尋的查詢 (query), 因此將線上商業意圖偵測議題分成網頁商業意圖偵測 (Web Page Online Commercial Intention Detection) </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/6707712430521415196/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=6707712430521415196' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/6707712430521415196'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/6707712430521415196'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2009/08/detecting-online-commercial-intention.html' title='Detecting Online Commercial Intention (OCI)'/><author><name>Stanley Fan 范登凱</name><uri>http://www.blogger.com/profile/00552079070299670477</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-226657832712117235</id><published>2009-08-12T05:28:00.004+08:00</published><updated>2009-08-24T09:38:33.578+08:00</updated><title type='text'>徵博士後研究及專任研究助理</title><summary type='text'>【職務說明】本校為執行「培育優質人力促進就業計畫－教育部補助大專校院研究人才延攬方案」，遴聘相關人員。相關權利義務，以雙方訂定之契約規範之。本方案延聘之研究人才不得兼任其他工作。本方案以95-97學年度畢業之本國籍.碩博士生為優先延聘對象。本方案不得聘任因本方案之實施而辦理休退學者。聘任單位/計畫主持人：國立中央大學資工系/張嘉惠　網路智慧暨資料探勘研究室【職務類別】博士後研究員1名，碩士或大專專任研究助理1名【工作範疇】主要研究內容：Web資料擷取及整合，機器學習及資料探勘，文件探勘及摘要協助研究計畫之進行、程式與報告撰寫【徵求條件】資工相關科系畢業，或對資料，文件，Web分析有經驗，且具有研究熱忱者。熟悉資料庫管理程式(mySQL&amp;php)Web伺服器架設熟悉Google , Yahoo 的 API and Widgets tools 的設計熟悉資料探勘及機器學習演算法的應用【</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/226657832712117235/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=226657832712117235' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/226657832712117235'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/226657832712117235'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/12/blog-post.html' title='徵博士後研究及專任研究助理'/><author><name>Jahui</name><uri>http://www.blogger.com/profile/04407009593178832508</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-3448462803609943071</id><published>2009-04-22T22:26:00.003+08:00</published><updated>2009-04-22T22:28:48.815+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='AI'/><category scheme='http://www.blogger.com/atom/ns#' term='Path Detection'/><category scheme='http://www.blogger.com/atom/ns#' term='Don'/><category scheme='http://www.blogger.com/atom/ns#' term='GPS'/><category scheme='http://www.blogger.com/atom/ns#' term='DBN'/><title type='text'>090423 AI Ch15 Probabilistic Reasoning Over Time</title><summary type='text'>Ai Ch15 Probabilistic Reasoning Over TimeView more presentations from Don.TW.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/3448462803609943071/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=3448462803609943071' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3448462803609943071'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3448462803609943071'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2009/04/090423-ai-ch15-probabilistic-reasoning.html' title='090423 AI Ch15 Probabilistic Reasoning Over Time'/><author><name>Don</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-5245645539062288928</id><published>2009-04-14T10:46:00.001+08:00</published><updated>2009-04-14T10:48:11.879+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='Schema matching'/><title type='text'>2009.04.13 Integrating Web Query Results: Holistic Schema Matching</title><summary type='text'>Integrating web query results: holistic schema matchingFull text                                                                          Pdf           (368 KB)                                                    Source                                                Conference on Information and Knowledge Management這次報告的主題是schema matching，而這篇的主題不是提出一個新的方法而是在現有的方法作效能的改善別且希望可以多到多來源網站的schema matching</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/5245645539062288928/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=5245645539062288928' title='1 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5245645539062288928'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5245645539062288928'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2009/04/20090413-integrating-web-query-results.html' title='2009.04.13 Integrating Web Query Results: Holistic Schema Matching'/><author><name>Emily Lin</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>1</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-3284190564058945539</id><published>2009-02-18T21:04:00.007+08:00</published><updated>2009-02-18T21:42:10.603+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='document classification'/><title type='text'>Emotion Classification Using Massive Examples Extracted from the Web</title><summary type='text'>本篇論文出自於COLING'2008, 其目的在解情緒分類問題 (emotion classification problem),主要方法係將大量的網路資源視為訓練資料進行分類器製作,並將情緒分類議題分解成二階段處理: sentiment classification and emotion classification, 實驗顯示透過兩階段分類的效果比起一階段分類方法效果顯著.第一階段處理(sentiment classification):採用SVM分類器及特徵集(n-gram and sentiment polarity words),訓練資料則是透過事先定義好的10個情緒類別字及其衍生情緒字詞當作是seed terms,並蒐集網路上相關的句子(稱為emotion provoking event)成為訓練資料集.第二階段處裡 (emotion classification):採用</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/3284190564058945539/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=3284190564058945539' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3284190564058945539'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3284190564058945539'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2009/02/emotion-classification-using-massive.html' title='Emotion Classification Using Massive Examples Extracted from the Web'/><author><name>Stanley Fan 范登凱</name><uri>http://www.blogger.com/profile/00552079070299670477</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-1517666217321420318</id><published>2008-12-29T17:01:00.002+08:00</published><updated>2008-12-29T17:22:02.970+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='web data extraction'/><title type='text'>DEByE─Data Extraction By Example</title><summary type='text'>以下是我對這篇論文的摘要：網路的蓬勃發展，使得越來越多使用者透過網路找尋有興趣的資訊，但如何擷取出特定的資訊是一件即為複雜的任務。因此本篇論文呈現如何從眾多的網頁資源當中去擷取資料的方法，使用者透過GUI介面提供自己想要擷取的例子(Example)，而不需要了解整個網頁架構為何，利用這些例子從新的網頁中去擷取資訊並產生pattern，另外DEByE也提出從下至上的程序來做資料的擷取，結果顯示出可利用較少的例子得到較好的擷取資訊效果。投影片：DEByE─Data Extraction By ExampleView SlideShare presentation or Upload your own. (tags: data &amp;)</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/1517666217321420318/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=1517666217321420318' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1517666217321420318'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1517666217321420318'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/12/debyedata-extraction-by-example.html' title='DEByE─Data Extraction By Example'/><author><name>swing</name><uri>http://www.blogger.com/profile/10386401967984346262</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-4495187623646448188</id><published>2008-12-29T03:07:00.002+08:00</published><updated>2008-12-29T03:09:59.034+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='web data extraction'/><title type='text'>FivaTech：Page-Level Web Data Extraction from Template Pages</title><summary type='text'>這一次閱讀的論文是以前Lab中博班學長與老師所發表的，目前Lab中的研究還有使用到以本篇論文技術為基礎的系統，但是仍然有些問題需要解決。如果這個系統能夠維護好並擴展改進的話，對未來網頁資料擷取的許多應用相信將很有幫助。最早Web IE採用的是supervised的方式，必須依靠人工標註才能產生擷取資料的規則。後來Dynamic Web的研究往Template-based的推論方向發展，產生了一些unsupervised方式的論文。FivaTech就是其中之一，它可以透過多個擁有相同Template的頁面來推論出其結構，也因此把這樣的方法歸類為Page-Level。FivaTech以多個網頁作輸入，最後產生共同的Template、Schema及個別的擷取資料。Template及Schema是以樹狀表示，與網頁的DOM Tree形成較一致的表現，其核心過程主要分成兩個較大的部份：Tree </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/4495187623646448188/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=4495187623646448188' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/4495187623646448188'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/4495187623646448188'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/12/fivatechpage-level-web-data-extraction.html' title='FivaTech：Page-Level Web Data Extraction from Template Pages'/><author><name>哲民</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-8666857662751997947</id><published>2008-12-24T06:06:00.017+08:00</published><updated>2008-12-24T06:49:26.844+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='cfp'/><title type='text'>Conference Call For Paper</title><summary type='text'>ICML 2009 Homepage and Paper Submission Website      Important Dates             January 26, 2009    Full paper submissions due (no separate abstract date)            Febrary 27, 2009    First round reviews available            March 10, 2009    Author response due            April 6, 2009    Acceptance notification            April 20, 2009    Final camera-ready version due            June 14, </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/8666857662751997947/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=8666857662751997947' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8666857662751997947'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8666857662751997947'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/12/cfp-icml-2009.html' title='Conference Call For Paper'/><author><name>Jahui</name><uri>http://www.blogger.com/profile/04407009593178832508</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-1581267280500648965</id><published>2008-12-23T13:30:00.003+08:00</published><updated>2008-12-23T18:29:09.789+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='Web application development'/><title type='text'>CompoWeb: A Component-Oriented Web Architecture</title><summary type='text'>這篇出自於www '08，是由亞洲微軟研究中心合發的一篇論文，在這邊他們提出舊有web application的開發方式缺點，並提出把component-based software development應用到web application development上面。他們提出了兩個新的concept，一個就是Gadget，另一個為Interface，前者就是如同一個COM元件一樣，主要是要呈現某些功能或是邏輯；後者是允許Gadget可以輸出一些PME model，而Interface就是來管理這些資訊，讓其他Gadget可以去使用這些PME。此研究如果可以開發出完整的開發平台，那必定是會造福很多開發者在管理Web application的程式碼更為容易，而且讓Web application更穩定以及提高安全性。2008.12.23 CompoWebView SlideShare </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/1581267280500648965/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=1581267280500648965' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1581267280500648965'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1581267280500648965'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/12/compoweb-component-oriented-web.html' title='CompoWeb: A Component-Oriented Web Architecture'/><author><name>Emily Lin</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-7035981542560391319</id><published>2008-12-05T16:23:00.000+08:00</published><updated>2008-12-05T16:26:07.054+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='social network'/><title type='text'>iLink : Search and Routing in Social Networks</title><summary type='text'>I Link Search And Routing In Social NetworksView SlideShare presentation or Upload your own.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/7035981542560391319/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=7035981542560391319' title='1 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7035981542560391319'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7035981542560391319'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/12/ilink-search-and-routing-in-social.html' title='iLink : Search and Routing in Social Networks'/><author><name>偽＊Hi Me</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>1</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-8323119933655991863</id><published>2008-12-02T22:11:00.001+08:00</published><updated>2008-12-02T22:12:55.769+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='GPS'/><title type='text'>Building Personal Maps from GPS Data</title><summary type='text'>081125 Building Personal Maps From Gps DataView SlideShare presentation or Upload your own. (tags: gps)</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/8323119933655991863/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=8323119933655991863' title='1 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8323119933655991863'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8323119933655991863'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/12/building-personal-maps-from-gps-data.html' title='Building Personal Maps from GPS Data'/><author><name>Don</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>1</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-72945653138663236</id><published>2008-11-17T14:26:00.001+08:00</published><updated>2008-11-17T15:33:38.405+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='social network'/><category scheme='http://www.blogger.com/atom/ns#' term='document classification'/><title type='text'>Learning Social Networks from Web Documents Using Support Vector Classifiers</title><summary type='text'>Learning Social Networks from Web Documents Using Support Vector Classifiers出自於IEEE Web Intelligence 2006, 本篇論文主要係透過機器學習的方式自動地去建立social network, 本研究首先假設已經存有不完整關聯 (incomplete relationship), 再透過SVM建立出完整的social network, 其中屬性的建立則是藉由網路文章 (web documents)來產生文件向量. 不難發現作者將判斷social network relationships的問題轉化成傳統的文件分類問題 (text classification problem), 所採用的判斷方法 (亦可視為分類方法) 則是SVM. 此外由於relationships在social </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/72945653138663236/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=72945653138663236' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/72945653138663236'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/72945653138663236'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/11/learning-social-networks-from-web.html' title='Learning Social Networks from Web Documents Using Support Vector Classifiers'/><author><name>Stanley Fan 范登凱</name><uri>http://www.blogger.com/profile/00552079070299670477</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-9208762744089171158</id><published>2008-11-05T22:02:00.003+08:00</published><updated>2008-11-05T22:06:22.210+08:00</updated><title type='text'>10/21 Data Selection for Support Vector Machine Classifiers</title><summary type='text'>摘要:       本論文介紹MSVM(Minimal Support Vector Machine)分類器，其概念為"基於SVM的架構下，減少其support vectors"。此分類器應用於Fraud detection等含有數以萬計的data　points，亦可以增進其他需要大量support vector才能決定的分類器之效能。　　       內文首先介紹SVM的作用及原理，其後介紹MSVM。此技術使用fast linear programming並加入了error term來減少所使用的vectors。最後提出了SLA(Successive Linearization Algorithm)的演算法來實作MSVM，最後是實驗比較。       將MSVM和FSV以及1-norm SVM對於七個資料集運算的結果做比較，明顯的發現MSVM所使用的support vectors</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/9208762744089171158/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=9208762744089171158' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/9208762744089171158'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/9208762744089171158'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/11/1021-data-selection-for-support-vector.html' title='10/21 Data Selection for Support Vector Machine Classifiers'/><author><name>Kuan-Chi-I</name><uri>http://www.blogger.com/profile/00305559112409878282</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-1778970156453603939</id><published>2008-11-02T22:23:00.010+08:00</published><updated>2008-11-11T13:16:51.800+08:00</updated><title type='text'>11/11 電腦鑑識程序之研究 A Survey of the Procedure for Computer Forensics </title><summary type='text'>文港主要報告在碩士班的研究，和老師以及同學報告，與否能在資料庫研究上延續，還請老師以及同學多多指教。摘要隨著資訊科技的興起與電腦時代的來臨，利用電腦以及網路犯罪的問題，讓執法單位面臨了更大的挑戰與困難，如何利用在電腦上的鑑識工具來取得有效的數位證據，是當前迫切所需要的課題，一般鑑識單位針對無法開機鑑識的電腦主機，大多使用Linux Live CD整合鑑識工具研究，但以目前電腦作業系統來看，以XP作業系統佔有率最高，本文研究提出電腦鑑識程序，利用XP Live CD整合電腦鑑識工具，建置實驗環境，針對電腦犯罪情事，採集相關證據，探討解決方法。通常單一鑑識工具所提出的證據不足，但以測試多套鑑識工具使用，過程中產生的相關癥結，最後彙整鑑識報告，相信能讓調查鑑識人員於數位證據擷取上趨於完善，在法庭呈現上的證據多一分效力。關鍵字：電腦鑑識、數位證據、Live CD。電腦鑑識程序之研究</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/1778970156453603939/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=1778970156453603939' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1778970156453603939'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1778970156453603939'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/11/1104-survey-of-procedure-for-computer.html' title='11/11 電腦鑑識程序之研究 A Survey of the Procedure for Computer Forensics '/><author><name>whukong</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-1420681080525885005</id><published>2008-10-27T17:34:00.004+08:00</published><updated>2008-11-04T21:43:20.908+08:00</updated><title type='text'>10/28 Regular Meeting: Monority Report in Fraud Detection Classification of Skewed Data</title><summary type='text'>摘要：本篇論文提出一個新的詐欺偵測方法，根據現存的詐欺偵測系統以及minority report,處理與skewed data有關的資料探勘問題。本篇論文提出的方法使用Backpropagation (BP)，Naive Bayesian (NB)，及C4.5演算法，配合over-sampling的方式。獨特的方在於使用一個meta-classifier (stacking)，選擇這些比較performance好的base classifiers，再將這些base classifiers的預測合併 (bagging)，用以改善cost savings (stacking-bagging)。實驗的結果證明stacking-bagging的performance比起傳統的algorithms要來的好。接著，本篇論文比較新的詐欺偵測方法與C4.5使用undersampling、</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/1420681080525885005/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=1420681080525885005' title='1 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1420681080525885005'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1420681080525885005'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/10/1028-regular-meeting-monority-report-in.html' title='10/28 Regular Meeting: Monority Report in Fraud Detection Classification of Skewed Data'/><author><name>萍華</name><uri>http://www.blogger.com/profile/13118754362107970593</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>1</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-3067090814566479878</id><published>2008-10-12T22:38:00.005+08:00</published><updated>2008-10-15T21:27:07.363+08:00</updated><title type='text'>10/14 Regular Meeting: Pattern Mining to Chinese Unknown Word Extraction</title><summary type='text'>本次報告為我的論文: Pattern Mining to Chinese Unknown Word Extraction。摘要如下:中文的文件資訊處理，由於沒有如歐美語系中有區隔符號(ex:空白)斷開每個辭彙，會遇到兩個大問題: 歧義性 與 未知詞問題。本篇的論文主要為解決中文的未知詞問題。未知詞又稱為OOV words(Out-Of-Vocabulary)，顧名思義，就是字典無法辨識的辭彙。由字典輔助的初步斷詞，會將這些不存在於字典裡的辭彙，錯誤的斷開成多個部份，如人名-王小明會被斷成 王 小 明 三個字。未知詞的任務，就是針對錯誤切割的多個字元部份，重新結合成一個正確的辭彙。本篇論文的架構，主要分成兩個階段:1. 第一階段為未知詞偵測部份。文章裡並非所有的字元都是未知詞的候選字元，必須先經過判斷，哪些是屬於未知詞的可能字元。因此一開始我們會找尋可能的未知詞字元，並著重於單音節字元(</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/3067090814566479878/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=3067090814566479878' title='1 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3067090814566479878'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3067090814566479878'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/10/1014-regular-meeting-pattern-mining-to.html' title='10/14 Regular Meeting: Pattern Mining to Chinese Unknown Word Extraction'/><author><name>totti-yang</name><uri>http://www.blogger.com/profile/02337865215775652142</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>1</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-2773613222744141150</id><published>2008-10-05T16:17:00.006+08:00</published><updated>2008-10-09T13:37:41.856+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='information extraction'/><title type='text'>10/07 Regular Meeting: A fuzzy symbolic Inference System for Postal Address Component Extraction and labeling</title><summary type='text'>論文連結：請點此論文摘要：    此篇論文主要是提出一個系統可以將地址的各項元素標示上不同的標籤，例如：地區名稱和街道號碼等資訊。對於有規律架構的地址格式和非標準架構的地址格式，後者在研究上顯得較不容易，因此本論文主要在解決unstructured address的問題。    首先，利用符號表示法來表示地址，建立知識庫作為標籤的依據，系統主要從輸入地址的各項元素和知識庫之間找出相似度，再利用提出的方法論來確定地址各項元素的標籤為何，所以fuzzy symbolic inference system即是對於一些較為模糊意義的地址元素加以判斷、擷取並標上標籤。論文介紹：a fuzzy symbolic system是這篇論文提出的系統，主要在解決unstrctured address的問題。一般我們所建的地址格式通常較為精確，可標示在Google Map上，但文中提到，類似India</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/2773613222744141150/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=2773613222744141150' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2773613222744141150'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2773613222744141150'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/10/1007-regular-meeting-fuzzy-symbolic.html' title='10/07 Regular Meeting: A fuzzy symbolic Inference System for Postal Address Component Extraction and labeling'/><author><name>swing</name><uri>http://www.blogger.com/profile/10386401967984346262</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-5517887655025201296</id><published>2008-10-02T18:54:00.001+08:00</published><updated>2008-10-03T13:56:04.806+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='Data integration'/><title type='text'>080930 Rugular Meeting：Relations, Cards, and Search Templates:User-Guided Web Data Integration and Layout</title><summary type='text'>這次報告的論文出自於UIST 2007，是由微軟、adobe及華盛頓大學所合作的一篇論文。     對現在人們來說，WWW是一個很大的資訊來源，使用者可以透過網路作旅遊計畫、購物、學習新事物或者是看電視等等，但是資訊內容極多，使用者有時候要找到適當的資訊會變的越來越難，所以他們的研究是提出一個互動式介面來幫助使用者更方便更容易的去收集、管理、組織及分享他們的資料。對於此研究，他們提出了三個技術，分別是：1. an interaction technique：允許使用者去規劃網站之間的關係並且使用這些關係自動地從多個網站收集內容。2. an interface：此介面是為了合併來自數個網站的內容並且以視覺化來呈現它們，此介面稱為card.3.a novel search paradigm：利用 search template 來從網路收集內容。以上三種技術是會建立在summaries </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/5517887655025201296/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=5517887655025201296' title='1 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5517887655025201296'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5517887655025201296'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/10/080930-rugular-meetingrelations-cards.html' title='080930 Rugular Meeting：Relations, Cards, and Search Templates:User-Guided Web Data Integration and Layout'/><author><name>Emily Lin</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>1</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-7653565994623829767</id><published>2008-09-30T11:56:00.002+08:00</published><updated>2008-09-30T19:02:13.965+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='ontology'/><category scheme='http://www.blogger.com/atom/ns#' term='information extraction'/><title type='text'>09/19 Regular Meeting：Timely Ontologies for Business Relations</title><summary type='text'>這一次報告的paper標題是"TOB：Timely Ontologies for Business Model"，出自WebDB 2008。主要是關於商業領域的資訊擷取應用，但是裡面所提到的方式其實也可以用在其它領域上。我覺得這種應用蠻有趣的，而且內容並不難懂，又可以了解其它種類的資訊擷取技術。底下是這篇paper的摘要部份：本體論(Ontology)一語出自哲學領域，主要探討存在的本質。近年來電腦科學家也將其應用在知識表達上，作為描述實體的概念以及實體間關係的模型。本篇作者主要提出一套法方來建立商業關係的本體論模型─TOB，此模型特色是包含了時間的因果關係。這種模型在商業智慧的應用上是很有用的，舉例來說，我們可以問像這一類的問題：微軟公司現在的執行長是誰？Google收購Youtube之後的獲利表現如何？TOB是基於YAGO模型[1]之上，加入了時間範圍的表達。在商業實體關係的擷取中，</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/7653565994623829767/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=7653565994623829767' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7653565994623829767'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7653565994623829767'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/09/0919-regular-meeting.html' title='09/19 Regular Meeting：Timely Ontologies for Business Relations'/><author><name>哲民</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-7987959538158826716</id><published>2008-09-17T12:32:00.005+08:00</published><updated>2008-09-17T12:39:18.318+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='Don'/><category scheme='http://www.blogger.com/atom/ns#' term='GPS'/><title type='text'>080902 Inferring High-Level Behavior from Low-Level Sensors</title><summary type='text'>低價GPS (Global Positioning System)技術的進展，引領了使用者定位的大量商業應用。 本篇是Learning and Inferring Transportation Routines之參考文獻。由於該篇的paremeter learning部份，講得不多，故找本篇來研讀。本篇的主題是，探討如何根據原始的GPS資料，來預測使用者的日常移動行為。本篇的想法較為原始，並未考慮goal與nolvety之情形。  作者使用的particle filter，是Bayesian filter的一種變形。particle filter以近似的方式來加速計算效率。作者並推導了EM演算法的步驟。  實驗的部份，資料來源為一位使用者的90天GPS資料。GPS訊號發送之時間間隔為2-10秒。實驗顯示，與之間的模型相比，預測的準確度有顯著提昇。Donald J. Patterson, </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/7987959538158826716/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=7987959538158826716' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7987959538158826716'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7987959538158826716'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/09/080902-inferring-high-level-behavior.html' title='080902 Inferring High-Level Behavior from Low-Level Sensors'/><author><name>Don</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-5309491815513602579</id><published>2008-08-27T20:33:00.002+08:00</published><updated>2008-08-27T20:47:32.536+08:00</updated><title type='text'>8/26 Regular Meeting: Learning On the Border:Active Learning in Imbalanced Classification data</title><summary type='text'>本篇論文著重於某些機器學習演算法，對於資料不平衡(imbalance data)處理效能的問題。在現實世界中一些需要分類的工作，比方說文字分類、詐欺偵測、醫療判斷等等，若以機器學習的方式處理，該怎樣達到最好的分類效果呢？標準的機器學習演算法對於平衡資料的處理可以達到極佳的預測效果，但是對於不平衡資料的預測效果卻是沒辦法達到要求。本篇論文解釋何謂資料不平衡，以實驗驗證Active Learning對於class imbalance可以達到不錯的效果，其次提出另一種有效率的方法--以更小的sample pool訓練資料，可以減少計算的時間、減少記憶體的消耗。Learning On The Border:Active Learning in Imbalanced classification DataView SlideShare presentation or Upload your own.</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/5309491815513602579/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=5309491815513602579' title='1 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5309491815513602579'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5309491815513602579'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/08/826-regular-meeting-learning-on.html' title='8/26 Regular Meeting: Learning On the Border:Active Learning in Imbalanced Classification data'/><author><name>萍華</name><uri>http://www.blogger.com/profile/13118754362107970593</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>1</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-737942259064973902</id><published>2008-08-26T17:04:00.000+08:00</published><updated>2008-09-26T17:10:48.285+08:00</updated><title type='text'>專題討論時間</title><summary type='text'></summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/737942259064973902/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=737942259064973902' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/737942259064973902'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/737942259064973902'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/09/blog-post.html' title='專題討論時間'/><author><name>Jahui</name><uri>http://www.blogger.com/profile/04407009593178832508</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-7071284433612336630</id><published>2008-08-18T21:06:00.004+08:00</published><updated>2008-08-19T23:57:53.484+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='text summarization'/><category scheme='http://www.blogger.com/atom/ns#' term='information extraction'/><category scheme='http://www.blogger.com/atom/ns#' term='document classification'/><title type='text'>8/20 Regular Meeting: Just-In-Time Contextual Advertising</title><summary type='text'>Just-In-Time Contextual Advertising, CIKM’2007論文和自身的研究有著高度的相關性, 目的同樣都是去解內容廣告配適問題 (Contextual Advertising).此論文考量到目前的網頁可分為靜態網頁 (static)和動態網頁 (dynamic), 在內容廣告配適技術上可針對靜態網頁做事前的內容分析(offline content analysis), 然而在面臨動態網頁時, 則需透過即時內容分析(online content analysis)方法. 因此該篇論文主要是探討如何在及時的情況下, 快速地針對網頁(動態和靜態)進行廣告配適. 為達到real-time 的廣告配適, 勢必要考量到網頁內容的大小, 如果分析該網頁全部的內容, 則會犧牲了網路傳輸和分析時間; 反之如果僅考量網頁的部分內容, 則可能會因為語意的缺乏, </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/7071284433612336630/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=7071284433612336630' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7071284433612336630'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7071284433612336630'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/08/820-regular-meeting-just-in-time.html' title='8/20 Regular Meeting: Just-In-Time Contextual Advertising'/><author><name>Stanley Fan 范登凱</name><uri>http://www.blogger.com/profile/00552079070299670477</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-7723815777443511574</id><published>2008-08-15T15:30:00.003+08:00</published><updated>2008-08-15T16:26:04.923+08:00</updated><title type='text'>08/12 Regular Meeting: Pattern Mining to Unknown Word Extraction</title><summary type='text'>本次報告為我的論文: Pattern Mining to Unknown Word Extraction。摘要如下:中文的文件資訊處理，由於沒有如歐美語系中有區隔符號(ex:空白)斷開每個辭彙，會遇到兩個大問題: 歧義性 與 未知詞問題。本篇的論文主要為解決中文的未知詞問題。未知詞又稱為OOV words(Out-Of-Vocabulary)，顧名思義，就是字典無法辨識的辭彙。由字典輔助的初步斷詞，會將這些不存在於字典裡的辭彙，錯誤的斷開成多個部份，如人名-王小明會被斷成 王 小 明 三個字。未知詞的任務，就是針對錯誤切割的多個字元部份，重新結合成一個正確的辭彙。本篇論文的架構，主要分成兩個階段:1. 第一階段為未知詞偵測部份。文章裡並非所有的字元都是未知詞的候選字元，必須先經過判斷，哪些是屬於未知詞的可能字元。因此我們一開始會找尋可能的未知詞字元，並著重於單音節(詞長為1)字元(96%</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/7723815777443511574/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=7723815777443511574' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7723815777443511574'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7723815777443511574'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/08/0812-regular-meeting-pattern-mining-to.html' title='08/12 Regular Meeting: Pattern Mining to Unknown Word Extraction'/><author><name>totti-yang</name><uri>http://www.blogger.com/profile/02337865215775652142</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-6181680353560787313</id><published>2008-08-04T14:20:00.001+08:00</published><updated>2008-08-04T14:21:33.121+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='social network'/><title type='text'>Tag-based Social Interest Discovery</title><summary type='text'>對於社群網站中常見的元素─標籤，我總覺得應該可以有更多利用的方法。我也曾經思考過標籤可以有哪些用途，甚至也想過可能可以代表使用者的興趣取向。可是不知道怎樣去分析，也不知道資料該如何取得。剛好WWW 2008有這一篇題為"Tag-based Social Interest Discovery"的paper，所以我就研讀了一下，學習別人的思考方式。這篇paper是由Yahoo公司的研究員所發表的，由於剛好收購del.icio.us網站，所以不難想像完全採用該網站的資料作分析與研究。底下是摘要部份：在Web 2.0的概念下，許多社群網站開始發展且越來越受到歡迎。其中主要區分成兩類，一類是以人為核心，如Facebook、MySpace等等；一類是以物件為核心，如YouTube、Flickr與del.icio.us等等。對於社群網站來說，發現使用者群體中的共同喜好是很有用的。</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/6181680353560787313/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=6181680353560787313' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/6181680353560787313'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/6181680353560787313'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/08/tag-based-social-interest-discovery.html' title='Tag-based Social Interest Discovery'/><author><name>哲民</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-3849506295349828932</id><published>2008-08-04T12:08:00.001+08:00</published><updated>2008-08-04T12:11:55.003+08:00</updated><title type='text'>Building Data Integration Queries by Demonstration</title><summary type='text'>本篇論文的標題為Building Data Integration Queries by Demonstration，出自IUI 2007。隨著網路越來越進步，許多資訊都可以從網路中取得。但使用者的資訊需求，常常是分佈於不同的網頁中。舉例來說，某個餐廳在餐廳評價網站中得到不錯的評價，但於衛生評定網站的評定是不衛生的，那麼對於想去此餐廳吃飯的人，可能就要去這些網站搜集、整理相關資訊以決定是否去此餐廳享用餐點。使用者想要從不同的網頁資料源取得資訊，大致有兩種方法。第一種是使用者自己至各個網站取得資料並整理，但這必需花費使用者許多時間。第二種是找尋資訊整合服務網站，利用網站的功能做資訊整合，不過這些網站所提供的資料來源都來自於固定某些網站，而使用者所需的資料，不一定是資訊整合網站有提供的。於是作者提出了一個名為Karma的系統，目標是方便使用者做不同網頁資料的整合，建立自己所需的mashup。</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/3849506295349828932/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=3849506295349828932' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3849506295349828932'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3849506295349828932'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/08/building-data-integration-queries-by.html' title='Building Data Integration Queries by Demonstration'/><author><name>ding</name><uri>http://www.blogger.com/profile/03674801346575090288</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-5162586826190504770</id><published>2008-06-19T18:30:00.001+08:00</published><updated>2008-06-19T18:32:53.375+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='google'/><title type='text'>Google 2008 Developer Day</title><summary type='text'>之前得知Google 2008 Developer Day在6/14舉行後，就線上報名參加了。雖然有把這個消息告知一些同學跟學長，最後發現卻只有我一個人報名。大概6/14這個時間非常接近期末考，大家都很忙。本來是想帶數位相機去照一些照片，無奈出發前發現電池竟然出問題，結果只好作罷。整個會議的時間從早上8:30報到開始到下午四點結束，大會地點在台北國際會議中心。由於開幕致詞時間在9:30，所以報到完後還有很多時間。在等待開始的期間，不經意看到旁邊一位老兄用apple的ibook上網在玩twitter，我猜他可能在抱怨怎麼還不開始。另外，現場休息區的佈置真的很Google的感覺，除了有一台wii跟xbox 360可以消磨時間外，還準備了一大堆的飲料跟零食。地上的椅子也都不太正常，有的是大汽球，有的是沙包，我看到有人坐在汽球上時還不小心跌倒。這一次大會主要分三個廳，主題都不太相同。上午的部份，</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/5162586826190504770/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=5162586826190504770' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5162586826190504770'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5162586826190504770'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/06/google-2008-developer-day.html' title='Google 2008 Developer Day'/><author><name>哲民</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-6071439533110626459</id><published>2008-05-19T13:41:00.010+08:00</published><updated>2008-06-05T11:24:54.901+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='information extraction'/><category scheme='http://www.blogger.com/atom/ns#' term='document classification'/><title type='text'>Automatic Identification of Pro and Con Reason in Online Reviews</title><summary type='text'>Automatic Identification of Pro and Con Reason in Online Reviews 這篇論文主要出自於COLING' 06,  其目的是將線上評價中的語句辨識出是否有含主觀意見, 進而將含主觀意見的句子分成pros 和 cons 兩個類別.因此, 給定一句使用者評論中的句子, 系統架構主要可以分為兩個階對進行, 分別為subjectivity identification and polarity classification phases:--subjectivity identification 將關於具有主觀意見的句子辨識出來.--polarity classification 將主觀意見的句子進行分類.subjectivity identification和polarity classification均採用supervised </summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/6071439533110626459'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/6071439533110626459'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/05/automatic-identification-of-pro-and-con.html' title='Automatic Identification of Pro and Con Reason in Online Reviews'/><author><name>Stanley Fan 范登凱</name><uri>http://www.blogger.com/profile/00552079070299670477</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-3748104378999511170</id><published>2008-05-07T17:21:00.001+08:00</published><updated>2008-05-07T17:24:34.010+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='mobile computing'/><title type='text'>Image Classification for Mobile Web Browsing</title><summary type='text'>這次報告的paper題目是"Image Classification for Mobile Web Browsing"，出處是"Proceedings of the 15th international conference on World Wide Web"。不難想像的是，作者是日本人，畢竟在日本，使用手機已經成為非常高頻率的一種活動。底下是摘要部份：對於只有小畫面的行動裝置使用者來說，瀏覽專為桌上型PC的大畫面所設計的網頁是不方便的。然而，隨著網路技術的提升與行動裝置的普及，這方面的需求也越來越多。目前已經有一些研究與商業產品正嘗試解決這方面的問題，其中，能夠正確地分辨網頁中image的種類是很有用的。舉例來說，去除網頁中某些image來簡化網頁內容以達到更符合小畫面瀏覽的程度。在這篇paper中，作者們將web imgaes分成11個種類。接著，從40個網站中收集到的3901個</summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3748104378999511170'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3748104378999511170'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/05/image-classification-for-mobile-web.html' title='Image Classification for Mobile Web Browsing'/><author><name>哲民</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-6143461766720015108</id><published>2008-03-02T21:00:00.003+08:00</published><updated>2008-03-02T21:05:12.980+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='web UI'/><title type='text'>OpenXUP─an Alternative Approach to Developing Highly Interactive Web Applications</title><summary type='text'>本篇paper出處是："Proceedings of the 6th international conference on Web engineering"，底下是摘要內容：製作更豐富與高互動性的Web Application需求益發增加，目前改善傳統的HTML方式的介面表示方法主要有兩種：一種是利用下載程式到browser中執行的方式如Java Applet或ActiveX；一種是近來很流行的AJAX。但是這兩種方式都有其缺點，前者當UI與程式功能比較複雜時，需要下載的程式碼也變得較多，並且由於程式功能在Client端執行，因此存在有安全性的風險；後者的UI則受限於Browser的JavaScript Engine與DHTML的表現能力。因此作者們提出了另一種可選擇的方式─OpenXUP，一種基於XUP(Extensible User Interface Protocol)的Web </summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/6143461766720015108'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/6143461766720015108'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/03/penxupan-alternative-approach-to.html' title='OpenXUP─an Alternative Approach to Developing Highly Interactive Web Applications'/><author><name>哲民</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-3711834998280319705</id><published>2008-01-03T18:06:00.000+08:00</published><updated>2008-01-03T18:18:03.039+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='information extraction'/><title type='text'>Data Integration Support for Mashups. &amp; A Framework for Rapid Integration of Presentation Components.</title><summary type='text'>Data Integration Support for Mashups這篇論文提出一個整合架構，讓使用者將現有的Services資料，透過作者提出的語法整合後轉為新的Service。作者以Online Citation Service為例，建構一個mashup，讓使用者輸入作者姓名，系統從DBLP得到和作者相關的論文，並且以這些論文資料去query Google Scholar，擷取論文的citation count。作者定義了Query、Fuse、Aggregate、Union等函式，幫助Mashup開發者透過這些函式組合出新元件。而在Query Google Scholar時，作者也設計3個heuristic方式以增加最後結果的正確性。A Framework for Rapid Integration of Presentation Components這篇論文提出了一個</summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3711834998280319705'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3711834998280319705'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/01/data-integration-support-for-mashups.html' title='Data Integration Support for Mashups. &amp; A Framework for Rapid Integration of Presentation Components.'/><author><name>tomelf</name><uri>http://www.blogger.com/profile/06082806571347157111</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-5519003812731842846</id><published>2008-01-03T14:18:00.000+08:00</published><updated>2008-01-04T14:08:53.115+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='social network'/><title type='text'>Mining Social Networks for Targeted Advertising</title><summary type='text'>本篇paper的題目是"Mining Social Networks for Targeted Advertising"，出處是："Proceedings of the 39th Annual Hawaii International Conference on System Sciences (HICSS'06)"。以下是本篇paper的摘要內容：在商業中，針對部份客戶作目標性的廣告推薦是很有用的。傳統上都是靠手動方式分析先前的歷史交易資料或是客戶的相關特徵，但是近年來隨著技術的進展，這部份已經開始利用自動化的工具來處理了。目前推薦系統產生目標廣告的技術主要有兩大類，一類是content-based，另一類則是social-based。前者主要是比對個人特徵與產品內容分類的匹配性，缺點是沒有利用到有影響力的其它人。後者則是利用客戶對產品的評等關係之間的關連來作推薦，</summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5519003812731842846'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5519003812731842846'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/01/mining-social-networks-for-targeted.html' title='Mining Social Networks for Targeted Advertising'/><author><name>哲民</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-1000394462175520228</id><published>2008-01-03T14:14:00.000+08:00</published><updated>2008-01-04T14:09:31.404+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='web modeling'/><title type='text'>A Framework for Rapid Integration of Presentation Components</title><summary type='text'>本篇paper的題目是 A Framework for Rapid Integration of Presentation Components，出處是Proceedings of the 16th international conference on World Wide Web WWW 2007，底下是我所寫的摘要部份：UI(User Inteface)的開發在軟體開發過程中是最費時的部份之一。尤其是在合成應用程式的軟體開發中，UI的可重複利用機制的需求變得越來越明顯。基於此，本篇paper的作者們提出了一個Presentation Integration Framework。透過這個Framework，可以讓開發人員利用現成的Web Application組合出一個新的混合程式，只需要提供適當的介面定義文件(以XPIL寫成)而不需撰寫Application間通訊的底層程式碼。</summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1000394462175520228'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1000394462175520228'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/01/framework-for-rapid-integration-of.html' title='A Framework for Rapid Integration of Presentation Components'/><author><name>哲民</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-8049921241404450014</id><published>2007-08-15T18:13:00.000+08:00</published><updated>2007-08-15T18:24:03.556+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='association rule'/><title type='text'>Mining Customer Value From Association Rules to Direct Marketing</title><summary type='text'>對於傳統的行銷方法而言，企業在決定行銷對象時，大多無特定目標發送廣告訊息。但事實上願意回應的消費者總是很少，造成企業在行銷資源上的浪費及顧客的反感。所以近年來，「Direct marketing for profit optimization」成為企業最想應用data mining技術解決的問題之一，它同時也是ACM在1998年所舉辦的KDD-CUP競賽問題。一般而言，企業必須在考慮每份廣告的行銷成本之下，找出潛在傾向回應的消費者，進而只針對該消費族群行銷，並使得企業獲得的利益最大化。然而，它通常會面臨「imbalanced data」及「inverse correlation」的挑戰。在一般的狀況下，付諸回應行動的消費者佔整體行銷名單的比例非常少(在KDD-CUP-98 dataset中只佔5%)，而大部分的data mining演算法卻是偏好學習資料整體的行為規則，</summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8049921241404450014'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8049921241404450014'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/08/mining-customer-value-from-association.html' title='Mining Customer Value From Association Rules to Direct Marketing'/><author><name>flyman</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-749415402662640863</id><published>2007-08-14T14:48:00.000+08:00</published><updated>2007-08-14T14:50:18.566+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='social network'/><title type='text'>POLYPHONET : An Advanced Social Network Extraction System from the Web</title><summary type='text'>本篇論文出自於WWW 2006，介紹的是一個人與人之間社群網路系統的建立，該系統是利用搜尋引擎，找出包含給定人名的網頁，並依擷取下來的網頁中人名的Co-occurrence，去計算人與人間的關聯度，進而建構出Social network。          論文當中，作者首先介紹一些在Social Network Extraction上過去常用的基本方法，並提及關於同名同姓的問題。接著，將人與人之間的關係分類成數種, 由於兩兩計算兩個人名之間的關聯度需要相當大的計算量、因此如何縮簡使用搜尋引擎query網頁的次數、也是重要的問題之一.  另外作者也提出以與人有關的word做為描述該人的metadata、以及利用metadata提供人與人間關聯度的另一種算法。          之後作者展示了POLYPHONET實際使用的介面與結果，且在最後提及Super Social Network </summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/749415402662640863'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/749415402662640863'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/08/polyphonet-advanced-social-network.html' title='POLYPHONET : An Advanced Social Network Extraction System from the Web'/><author><name>IYO</name><uri>http://www.blogger.com/profile/11251684987125700139</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-8845878118750859293</id><published>2007-08-09T17:34:00.000+08:00</published><updated>2007-08-09T17:39:17.723+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='information extraction'/><title type='text'>Homepage live: automatic block tracing for web personalization</title><summary type='text'>這篇論文觀察到personal homepage的趨勢，包括Microsoft和Google都有做personal homepage，並在homepage中讓人放置許多功能區塊。因此論文提出一個系統Homapge Live，可以讓使用者在這個網頁上建立個人首頁，並且加入小區塊。Homepage live的小區塊特色是可以讓使用者作網頁中的block tracing，也就是網頁區塊的monitering。這篇論文重點在於系統在目標網頁版本更新後，如何正確地抓取使用者所想看到的區塊。首先將原網頁P_old和新網頁P_new表示成DOM Tree然後執行以下步驟：找出Fix Nodes：Fix Node代表在P_old及P_new中內容重複的node。產生Reduced Tree：刪除掉P_old及P_new中的Fix Nodes。Mapping：此時P_old和P_new都已經成為</summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8845878118750859293'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8845878118750859293'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/08/homepage-live-automatic-block-tracing.html' title='Homepage live: automatic block tracing for web personalization'/><author><name>tomelf</name><uri>http://www.blogger.com/profile/06082806571347157111</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-701642341641941372</id><published>2007-08-07T15:02:00.006+08:00</published><updated>2008-08-11T16:11:35.646+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='aboutlab'/><title type='text'>研究計畫</title><summary type='text'>   計畫名稱         補助或委託機構         起訖年月         計畫內擔任的工作             行動關懷社會之建構與服務(3/3)         國科會(97-2627-E-008-001-)         2008/8/1 至 2009/7/31         共同主持人             線上拍賣網站中銷售策略的研究         國科會(97-2221-E-008-088-)         2008/8/1 至 2009/7/31         主持人             Web資訊整合服務系統開發之研究         國科會(96-2221-E-008-091-MY2)         2007/8/1 至 2009/7/31         主持人             行動關懷社會之建構與服務(2/3)</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/701642341641941372/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=701642341641941372' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/701642341641941372'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/701642341641941372'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/08/blog-post_07.html' title='研究計畫'/><author><name>Jahui</name><uri>http://www.blogger.com/profile/04407009593178832508</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-5192574573121198622</id><published>2007-08-07T14:48:00.003+08:00</published><updated>2010-06-06T21:08:04.626+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='aboutlab'/><title type='text'>論文發表 (updated 06/05/2010)</title><summary type='text'>Journal PapersT.-K. Fan and C.-H. Chang: Sentiment Oriented Contexture Advertising. To appear in Journal of Knowledge and Information System. M. Kayed and C.-H. Chang: FiVaTech: Page-Level Web Data Extraction from Template Pages.  IEEE Trans. Knowl. Data Eng. Vol. 22, No.2, pp. 249-263, 2010.  T.-K. Fan and C.-H. Chang: Exploring Evolutionary Technical Trends From Academic Research Papers.  J. </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/5192574573121198622/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=5192574573121198622' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5192574573121198622'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5192574573121198622'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/08/blog-post.html' title='論文發表 (updated 06/05/2010)'/><author><name>Jahui</name><uri>http://www.blogger.com/profile/04407009593178832508</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-7223005517230562084</id><published>2007-07-08T10:25:00.053+08:00</published><updated>2011-11-02T14:56:13.784+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='aboutlab'/><title type='text'>實驗室成員 (Updated 09/20/2011)</title><summary type='text'>PhD 100 張淵琮 (Yuan-Chung Chang)100 周建龍(Chien-Lung Chou)99 邱妮塔 (Pawar Nita) : pawar.nita@gmail.com98 巫孟倫 (Meng-Lun Wu) : alen16@db.csie.ncu.edu.twMaster100 陳貞伶(Chen-Ling Chen)100 林伯瀚 (Po-Han Lin)100 黄俞翔(Yu-Hsiang Huang)100 鄭乃洪(Zheng Nai Hong)100 陳慶治 (Qing-Zhi Chen)99 吳禹欣 (Yu-Hsin Wu), abc771003@gmail.com99 陳柏志 (Po-chih Chen), chenbojyh@gmail.com99 趙士賢 (Shi-Xian Zhao), kizu1124@gmail.com 99 吳文斌 (</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/7223005517230562084/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=7223005517230562084' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7223005517230562084'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7223005517230562084'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2008/07/blog-post.html' title='實驗室成員 (Updated 09/20/2011)'/><author><name>Jahui</name><uri>http://www.blogger.com/profile/04407009593178832508</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-3394330945496828915</id><published>2007-07-04T20:34:00.000+08:00</published><updated>2007-07-05T13:42:32.854+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='document clustering'/><title type='text'>A New Suffix Tree Similarity Measure for Document Clustering</title><summary type='text'>A New Suffix Tree Similarity Measure for Document Clustering 這篇論文想做的是關於文章的自動分群，使用的是結合Suffix Tree和Vector Space Document的方法，作者稱之為NSTC(new suffix tree document clustering algorithm)。  作者在論文中首先介紹了一個在1998年所提出的STC(suffix tree document clustering algorithm)，STC主要可分為三個步驟，第一是產生document的suffix tree，找出內部節點做為base cluster，第二步是base cluster 挑選，第三步是依據Jaccord coefficient來決定node是否具有direct link，最後各個connected graph</summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3394330945496828915'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/3394330945496828915'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/07/new-suffix-tree-similarity-measure-for.html' title='A New Suffix Tree Similarity Measure for Document Clustering'/><author><name>IYO</name><uri>http://www.blogger.com/profile/11251684987125700139</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-4109507229838404604</id><published>2007-07-02T14:40:00.000+08:00</published><updated>2007-07-02T14:47:00.915+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='information extraction'/><title type='text'>Supporting end-users in the creation of dependable web clips</title><summary type='text'>當使用者利用 Microsoft FrontPage 等Web Authoring Tools建立自己的網頁時，常希望加入可以對網路資訊作monitoring的小區塊，例如擷取天氣預報、新聞…等等，將個人資料作information integration。有些網站有自行提供API或Mashup等工具讓使用者做資料擷取，但許多網站沒有相關工具。Web Clip: 作者定義為end-user在自行製作網頁時，能夠即時擷取網頁資訊並顯示在網頁上的小區塊。一個Web Clip會針對某目標網站中的資料作monitoring，多個web clips可以整合到單一個人網頁。這篇論文提出Web Clipper，一個圖形化介面的Web Clip製作流程，讓使用者在 Microsoft Frontpage 中透過簡單操作就可以快速產生Web Clip並加入個人網頁中，</summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/4109507229838404604'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/4109507229838404604'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/07/supporting-end-users-in-creation-of.html' title='Supporting end-users in the creation of dependable web clips'/><author><name>tomelf</name><uri>http://www.blogger.com/profile/06082806571347157111</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-2809617200469662522</id><published>2007-06-28T22:26:00.020+08:00</published><updated>2011-08-23T15:07:01.410+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='aboutlab'/><title type='text'>歷屆畢業論文</title><summary type='text'>PhD Thesis
98 范登凱  網路個人化廣告配置之研究 
94 黃國瑜 交易型資料庫之跨交易關聯規則探勘之研究 

Master Thesis

99-2劉睿哲 使用擴充資料進行共分群的協同式推薦系統 
99-2陳志銘 基於多元化部落格網頁之自動化擷取部落格主要文章 
99-2霍冠樺 手機廣告：使用者、廣告商及電信業者三贏架構 
99-2林書彥 形聲字發音規則探勘

98-2 林衍伶 非監督式網頁層次包覆程式之驗證 
98-2 張志豪（在職專班） 機器學習應用於樣版網頁擷取之研究 
98-2 潘立人（在職專班）資料搜尋系統視覺化與多維度分析之設計：以資訊工程研究論文檢索系統為例 


97-2 李淑瑩 英文郵政地址與鄰近相關資訊擷取之研究 
97-2 楊萍華 部落格意見檢索系統之設計-部落格內文之擷取與不相關部落格之過濾 

97-1 楊傑程  </summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/2809617200469662522/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=2809617200469662522' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2809617200469662522'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2809617200469662522'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/06/blog-post.html' title='歷屆畢業論文'/><author><name>Jahui</name><uri>http://www.blogger.com/profile/04407009593178832508</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-7896308273878246926</id><published>2007-06-05T13:44:00.000+08:00</published><updated>2007-06-07T17:09:23.223+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='query translation'/><title type='text'>A Study of Statistical Models for Query Translation ：Finding a Good Unit of Translation</title><summary type='text'>利用雙語字典來翻譯query，目前有兩個比較重要的課題：一是如何增強字典的覆蓋度，二是如何從字典中選出正確的字來進行翻譯。而本篇文章著重在後者。作者在論文中結合了三種的Model，分別為Co-occurence Model、Noun Phrase Model和Dependency Translation Model來幫助翻譯。這三種Model都是使用統計的資料（如：corpus），並藉由定義特徵函數，求最大化來取得較有可能的翻譯方式，三者的差異之處在於是針對不同的translation unit來進行處理。Noun Phrase Model針對的是名詞片語，其利用大部分名詞片語可以直譯的特性，配合由字典中來尋找翻譯的Template，以對名詞片語進行翻譯，希望最佳化Translation Template和word selection機率乘積所組成的特徵函數。Dependency </summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7896308273878246926'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/7896308273878246926'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/06/study-of-statistical-models-for-query.html' title='A Study of Statistical Models for Query Translation ：Finding a Good Unit of Translation'/><author><name>IYO</name><uri>http://www.blogger.com/profile/11251684987125700139</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-5784183978045252571</id><published>2007-06-05T12:52:00.002+08:00</published><updated>2008-08-07T14:47:19.524+08:00</updated><title type='text'>Digital Content Recommender on the Internet</title><summary type='text'>這篇論文的出處是從IEEE INTELLIGENT SYSTEMS 2006，目的是做數位文章的推薦。此系統分成development及online use兩個環境。development 環境做了內容跟使用者的分析並建立數個元件以供online use environment使用。   文章內容分析部分，先用貝氏分類器做文章分類，同時將information gain高的字放入word dictionary，之後一些base的建立都會參考到word dictionary。接下來用SOM做文章分群，並建立出article-keywords index跟keyword-articles index database。   使用者分析部分，依照使用者對每個之前所分群好的文章類別之RFM值用SOM做分群，並依每個群建出user preference rules。online use </summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5784183978045252571'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5784183978045252571'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/06/digital-content-recommender-on-internet.html' title='Digital Content Recommender on the Internet'/><author><name>ding</name><uri>http://www.blogger.com/profile/03674801346575090288</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-1174682222913297435</id><published>2007-06-04T11:44:00.000+08:00</published><updated>2007-06-04T12:02:44.417+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='information extraction'/><title type='text'>Interactive Wrapper Generation with Minimal User Effort</title><summary type='text'>這篇論文出自WWW 2006，目標是提出一個半自動化的wrapper generation方法。不同於其他semi-automatic和unsupervised wrapper，這篇論文並沒有很困難的方法，主要是提供一個使用者介面利用minimal user effort來完成wrapper工作。內容可以區分為兩個部分：Interactive User Interface：只需少量的training pages，讓使用者highlight頁面中的tuples，然後系統根據使用者highlight的部分做處理，將文件轉為DOM parse tree。Wrapper Generation System：系統定義了自己的Extraciton Language predicates，來作為產生extraction patterns的工具。首先透過training tuple產生dom_path</summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1174682222913297435'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1174682222913297435'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/06/interactive-wrapper-generation-with.html' title='Interactive Wrapper Generation with Minimal User Effort'/><author><name>tomelf</name><uri>http://www.blogger.com/profile/06082806571347157111</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><media:thumbnail xmlns:media='http://search.yahoo.com/mrss/' url='http://bp0.blogger.com/_mDIXmSDmhgE/RmOLUySgW-I/AAAAAAAAABo/iFj95yKpzg0/s72-c/%E5%9C%96%E7%89%871.jpg' height='72' width='72'/></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-8251417387690399890</id><published>2007-06-01T06:49:00.003+08:00</published><updated>2008-12-24T06:53:46.304+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='cfp'/><title type='text'>Conference Call For Paper</title><summary type='text'>ICDE 2008 Homepage and Submission WebsiteResearch and Industrial papers              Abstract deadline:       June 22, 2007 5pm PDT             Paper submission deadline:       June 27, 2007 5pm PDT           Notification:       October 12, 2007           Camera-ready deadline:      November 22, 2007 -</summary><link rel='replies' type='application/atom+xml' href='http://web204seminar.blogspot.com/feeds/8251417387690399890/comments/default' title='張貼意見'/><link rel='replies' type='text/html' href='http://www.blogger.com/comment.g?blogID=8830511825665412342&amp;postID=8251417387690399890' title='0 個意見'/><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8251417387690399890'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8251417387690399890'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/06/conference-call-for-paper.html' title='Conference Call For Paper'/><author><name>Jahui</name><uri>http://www.blogger.com/profile/04407009593178832508</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author><thr:total>0</thr:total></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-4492667895588453827</id><published>2007-05-24T18:27:00.008+08:00</published><updated>2008-08-07T15:06:07.781+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='aboutlab'/><title type='text'>實驗室成員</title><summary type='text'>Advisor張嘉惠(Chia-Hui Chang): http://jahuichang.blogspot.com/PhD Students范登凱(Stanley Fan): http://stanley-fan.blogspot.com/柯嘉穆(Mohammed Kayed):Master Students95 楊潔程 (Chieh-Cheng Yang) http://totti-yang.blogspot.com/95 丁昭廷 (Chao-Ting Ting) http://dingqoo.blogspot.com/95 蘇詠勝 (Yueng-Sheng Su) http://iyo-studio.blogspot.com/96 廖哲民 (Che-Min Liao) http://liouville.blogspot.com/96 楊萍華 (Ping-Hua Yang) http</summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/4492667895588453827'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/4492667895588453827'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/05/0516-ranking-objects-by-exploiting_24.html' title='實驗室成員'/><author><name>totti-yang</name><uri>http://www.blogger.com/profile/02337865215775652142</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-2630308789697535682</id><published>2007-05-16T17:58:00.001+08:00</published><updated>2007-06-05T07:32:02.553+08:00</updated><title type='text'>Ranking Objects by Exploiting Relationships:Computing Top-K over Aggregation</title><summary type='text'>這篇paper是來自於SIGMOD 2006，主要是探討給定一些關鍵字查詢時，所得到的查詢結果(文件)中的target object如何進行評分及排序，同時考慮在不需進行完整計算的可能性，因為我們只需要找出最好的前k項target object就足夠了。因此，這篇paper就是以上述為目標，並提出一個計算target object 分數，節省時間的演算法(called early termination)。論文中首先界定document與object，以及他們之間的關係。舉例來說，在找尋領域專家的應用上，我們可以輸入代表此領域的關鍵字，從而得到相關論文paper (document)，而我們想要找尋的target object，則是paper的author。document與target object間的包含關係，我們可以用一個table來記錄，也就是我們在某個document內發現哪些</summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2630308789697535682'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2630308789697535682'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/05/0516-ranking-objects-by-exploiting.html' title='Ranking Objects by Exploiting Relationships:Computing Top-K over Aggregation'/><author><name>totti-yang</name><uri>http://www.blogger.com/profile/02337865215775652142</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-5968504970346104264</id><published>2007-05-16T17:53:00.000+08:00</published><updated>2007-05-16T18:00:36.552+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='association rule'/><title type='text'>Using metarules to organize and group discovered association rules</title><summary type='text'>對於小型資料集來說，Association Mining Algorithm 能輕易地找出項目之間的關係，並產生關聯規則供使用者解讀。但若要處理高維的資料集時，產生的大量關聯規則通常會讓使用者無褔消受，這是因為高維的資料集容易產生許多reduntdant rule，而reduntdant rule通常起因於規則之間的overlap及containment，必須加以刪除。本篇論文針對我們所找出來的discovered rules，加以組織、分群，減少discovered rules數量，並以metarule來幫助使用者了解discovered rules彼此間的關係，以求降低使用者解讀discovered rules時的負擔。本篇論文提出的方法包含四個步驟：1. Finding meatrules所謂的metarules係由兩個所找出的association rules R1, R2</summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5968504970346104264'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5968504970346104264'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/05/using-metarules-to-organize-and-group.html' title='Using metarules to organize and group discovered association rules'/><author><name>flyman</name><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-1027063125902979770</id><published>2007-05-08T01:59:00.009+08:00</published><updated>2009-08-05T00:43:36.549+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='aboutlab'/><title type='text'>實驗室簡介</title><summary type='text'>Web智慧暨資料探勘實驗室本實驗室為中央大學資訊工程系張嘉惠教授所領導的研究團隊。實驗室位於工五館A棟307(分機35327)及研究中心二館206(分機57868)。實驗室成員目前有博士生二位，碩士生十位，碩士專班七位。實驗室的主要研究領域為全球資訊網及資料探勘兩大方向:Web由1.0發展到2.0，一個很大的差別在於網站從單純的提供使用者資訊的下載與閱覽，轉而提供使用者更多的上傳及分享的功能。現今的Web，每個人不僅是Web上的資訊消費者，同時也有能力藉由更方便的工具去創造資訊，我們可以說當今Web所包含的資料比過往更豐富完整。不過當我們將Web視為無所不包的資料庫，資訊的整合與利用就更為重要。近年來Mashup網站紛紛成立，正說明資訊整合的重要應用。不過再多的混搭網站也難顧到每個使用者的各別需求，因此以往被視為專業性操作的資訊整合，有必要提供更為公眾化的操作。過去我們實驗室即針對</summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1027063125902979770'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/1027063125902979770'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/05/blog-post.html' title='實驗室簡介'/><author><name>Jahui</name><uri>http://www.blogger.com/profile/04407009593178832508</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-2002251418250316942</id><published>2007-04-21T13:37:00.000+08:00</published><updated>2007-04-21T13:49:23.134+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='association rule'/><title type='text'>Lazy Associative Classification</title><summary type='text'>這篇paper主要是介紹了一個新的associative classification的作法。一般來說，associative classifier會比decision tree classifier的準確率要來得高。因為decision tree classifier 是用greedy(local) search的方法，選出目前擁有最高information gain的attribute，加進tree裡面形成新的split node，對於此node的子樹來說，皆為該attribute與其他所剩下的attribute中，挑選擁有最高information gain的attribute來繼續splitting，直到所有的instance屬於同一個class或是此node已低於某特定的minimum support threshold等等。decision tree classifier</summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2002251418250316942'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/2002251418250316942'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/04/lazy-associative-classification.html' title='Lazy Associative Classification'/><author><name>totti-yang</name><uri>http://www.blogger.com/profile/02337865215775652142</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-8294166007827644457</id><published>2007-04-11T10:35:00.000+08:00</published><updated>2007-04-18T03:51:47.612+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='text summarization'/><category scheme='http://www.blogger.com/atom/ns#' term='opinion extraction'/><title type='text'>Movie review mining and summarization</title><summary type='text'>這篇論文出自CIKM’06，目的是做movie review mining和summmarization，主要是藉由產生正面與負面的feature-opinion pairs做為此電影資訊的摘要。此篇論文提出的方法是一開始先利用WordNet、IMDB的movie casts和labeled training data去產生feature跟opinion的keyword list，利用此list找出句子中的feature words和opinion words且決定feature words是屬於哪個feature class和opinion words是正面的還負面的；訓練階段會利用pattern mining 找出dependency grammar graph中feature到opinion pair之間的常見pattern，做為測試階段feature-opinion pair</summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8294166007827644457'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/8294166007827644457'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/04/movie-review-mining-and-summarization.html' title='Movie review mining and summarization'/><author><name>ding</name><uri>http://www.blogger.com/profile/03674801346575090288</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-5499740935972521492</id><published>2007-04-11T09:28:00.001+08:00</published><updated>2007-04-18T03:52:41.828+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='keyword selection'/><category scheme='http://www.blogger.com/atom/ns#' term='document classification'/><title type='text'>A study on automatically extracted keywords in text categorization</title><summary type='text'>A study on automatically extracted keywords in text categorization本篇文章出自於ACL’2006, 其主要目的是探討 “Keyword” 對文件分類的影響力.keyword extraction: 列舉出幾個常用的方法, 例如: n-grams, PoS, and Chunking. 在feature value 指派的方面, 描述出幾個常用的方法, 例如: tf, tf*idf, relative position of the first occurrence, PoS tag, 其中PoS tag是作者在 “Improved Automatic Keyword Extraction Given More Linguistic Knowledge”所提出來的方法, 在得到input feature and </summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5499740935972521492'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/5499740935972521492'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/04/study-on-automatically-extracted.html' title='A study on automatically extracted keywords in text categorization'/><author><name>Stanley Fan 范登凱</name><uri>http://www.blogger.com/profile/00552079070299670477</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-4523329237293010045</id><published>2007-04-10T10:00:00.000+08:00</published><updated>2007-04-10T12:01:14.147+08:00</updated><title type='text'>Pollock: Automatic Generation of Virtual Web Services from Web Sites</title><summary type='text'>這一篇論文是出自ACM SAC 2005，主要目的是設計一個快速產生Web Service的方法。假設現在我們想要整合不同的網路書店，我們可能會先找100個書店，然後設法抓出網站中的拍賣資訊。此時有些書店有提供Web Service直接查詢，但有些書店則缺乏相關資訊。這篇論文的構想是為沒有Web Service的網站自動產生"Virtual Web Service"，如此一來便可以快速整合不同書店間的資訊。對於沒有Web Service的網站，首先會利用Wrapper技術擷取該網站的回傳網頁，在這裡使用XWRAP。XWRAP是一個以XML為基礎的wrapper，可 以將wrap出來的資訊轉為XML呈現。同時論文設計了一個translator，用來轉換SOAP和Html GET/POST的資訊，讓Wrapper執行時資訊可以順利傳遞。在Wrapper工作完成後，接下來產生Web </summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/4523329237293010045'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/4523329237293010045'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/04/pollock-automatic-generation-of-virtual.html' title='Pollock: Automatic Generation of Virtual Web Services from Web Sites'/><author><name>tomelf</name><uri>http://www.blogger.com/profile/06082806571347157111</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-872932255356999690</id><published>2007-03-02T09:44:00.000+08:00</published><updated>2007-03-13T04:47:40.678+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='recommendation'/><category scheme='http://www.blogger.com/atom/ns#' term='association rule'/><title type='text'>Adaptive-Support Association Rule Mining for Recommender Systems</title><summary type='text'>一般來說，我們在使用Association Rule Mining Algorithm時，例如Apriori，我們必須事先設定minimum support threshold。 但如何選擇適當的值卻不是那麼直覺，常常只能依靠暴力法來解決。此外，也因為沒有對規則有所限制，所以任何的frequent itemset都會產生規則，使得執行時間大幅增加。 上述以market basket analysis為精神的演算法不適用於推薦系統。因為推薦系統常對特定使用者進行推薦商品的動作，所以不須產生所有的frequent itemset。此外推薦系統必須考慮執行效率，特別是要online運作的規則。所以在這篇論文中，作者根據前人的CBA-RG演算法精神加以修改成適合應用於推薦系統的演算法－ASARM。ASARM不須事先指定minimum support threshold，</summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/872932255356999690'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/872932255356999690'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/03/adaptive-support-association-rule.html' title='Adaptive-Support Association Rule Mining for Recommender Systems'/><author><name>Jahui</name><uri>http://www.blogger.com/profile/04407009593178832508</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry><entry><id>tag:blogger.com,1999:blog-8830511825665412342.post-9036702492012172996</id><published>2007-02-14T17:03:00.000+08:00</published><updated>2007-03-22T14:30:29.770+08:00</updated><category scheme='http://www.blogger.com/atom/ns#' term='text summarization'/><title type='text'>Social Summarization of Online Auction Feedbacks</title><summary type='text'>這一篇IUI 2005的論文主要解的問題是拍賣網站上對於一個賣家評價的summarization．一般而言，我們在拍賣網站上想要評估一個賣家時，通常我們會先看其他人對這個賣家的評價，並且以多數人的意見做為此賣家評價摘要的主要內容，換句話說，意見的頻率是文件摘要的主要採用的條件。但是通常這樣的評價都是正面居多，很難得到一個比較真實的評價．因此作者提出先對同一個寫評價的使用者，將其（對於不同賣家）所有的意見整理出來，用以過濾禮貌性敍述（也就是那些對每個賣家都寫的意見），保留比較特別的意見，做為summarization的參考．比較有趣的是如何評估這篇所提的Social Summarization比一般的Summarization 好? 本篇作者以對不同賣家摘要所得的壓縮率標準差來說明，壓縮率標準差大表示所做的摘要較有鑑別度，而實驗結果也的確顯現這樣的現象．而處理這些feedback</summary><link rel='edit' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/9036702492012172996'/><link rel='self' type='application/atom+xml' href='http://www.blogger.com/feeds/8830511825665412342/posts/default/9036702492012172996'/><link rel='alternate' type='text/html' href='http://web204seminar.blogspot.com/2007/02/social-summarization-of-online-auction.html' title='Social Summarization of Online Auction Feedbacks'/><author><name>Jahui</name><uri>http://www.blogger.com/profile/04407009593178832508</uri><email>noreply@blogger.com</email><gd:image rel='http://schemas.google.com/g/2005#thumbnail' width='16' height='16' src='http://img2.blogblog.com/img/b16-rounded.gif'/></author></entry></feed>
