MISC

2012年12月5日

Classifying Twitter Users for Spatio-temporal Entity Retrieval

研究報告データベースシステム(DBS)
  • Liang Yan
  • ,
  • Qiang Ma
  • ,
  • Masatoshi Yoshikawa

2012
15
開始ページ
1
終了ページ
6
記述言語
英語
掲載種別

Spatio-temporal entity retrieval is a task for searching the entities with certain time and certain place, such as some commodities or events, from Twitter and Facebook, the social network with mass real-time update information. On Twitter, there are some users who tweet to the unspecified large number of other users, such as shops or local governments etc, while some other users who almost tweets to their friends. In this paper, we call the former as open account, while the latter as closed account. The expression in tweets and credibility of the two type of users can be different. For example, sometimes, an open account said "there are still some stock in the shop" about a commodity, while a closed account said "I didn't get it". In order to improve the accuracy of spatio-temporal ER, it is necessary to classify Twitter users. In this paper, we propose the method to classify Twitter users into open accounts and closed accounts. We use both the feature of user profile, such as address or telephone number etc. and the followers distribution. If the followers distribution is scattered, we treat it open account, while closed account otherwise.Spatio-temporal entity retrieval is a task for searching the entities with certain time and certain place, such as some commodities or events, from Twitter and Facebook, the social network with mass real-time update information. On Twitter, there are some users who tweet to the unspecified large number of other users, such as shops or local governments etc, while some other users who almost tweets to their friends. In this paper, we call the former as open account, while the latter as closed account. The expression in tweets and credibility of the two type of users can be different. For example, sometimes, an open account said "there are still some stock in the shop" about a commodity, while a closed account said "I didn't get it". In order to improve the accuracy of spatio-temporal ER, it is necessary to classify Twitter users. In this paper, we propose the method to classify Twitter users into open accounts and closed accounts. We use both the feature of user profile, such as address or telephone number etc. and the followers distribution. If the followers distribution is scattered, we treat it open account, while closed account otherwise.

リンク情報
CiNii Articles
http://ci.nii.ac.jp/naid/110009490413
CiNii Books
http://ci.nii.ac.jp/ncid/AN10112482
URL
http://id.nii.ac.jp/1001/00087373/
ID情報
  • CiNii Articles ID : 110009490413
  • CiNii Books ID : AN10112482

エクスポート
BibTeX RIS