東京大学大学院情報理工学系研究科電子情報学専攻
Department of Information and Communication Engineering, Graduate School of Information Science and Technology, The University of Tokyo
Abstract
Microblog, especially Twitter today has become an important tool to propagate public information among Internet users. The content of Twitter is an extraordinarily large number of small textual messages, posted by millions of users, at random or in response to perceived events or situations. However, messages of Twitter (tweets) cover so many categories including news, spam and others that it's difficult to provide public information directly. Since the traditional search cannot meet demands of tweets of some category, we aim to classify tweets automatically into defined categories to help users search. In our paper, we focus on approaches of collecting a corpus automatically for training classifiers. We proposed two approaches that are based on typical Twitter user accounts and based on Twitter lists using label propagation respectively. Using the corpora, we built classifiers, which are able to determine news, commercial and private tweets. Experiments evaluations show our proposed techniques are effective. In our search, we worked with Japanese, but the proposed approaches can be used with any other language.
発行年
2012-03-22
日本十進分類法
007
学位名
修士(情報理工学)
学位
master
研究科・専攻
情報理工学系研究科電子情報学専攻
学位授与年月日
2012-03-22
アイテムへのリンクの際はタイトル右肩にある「Permalink」をご利用ください！
Please use "Permalink" at the top right of the title when linking to items!
Classification of Microblog Posts Based on Information Publicness
48106424.pdf
(1.32MB)
