Authorship Identification for Hundred-thousand-scale Microblog Users in the Web

2016 Fiscal Year Final Research Report

• Project/Area Number: 25280113
• Research Category: Grant-in-Aid for Scientific Research (B)
• Allocation Type: Partial Multi-year Fund
• Section: 一般
• Research Field: Web informatics, Service informatics
• Research Institution: Waseda University
• Principal Investigator: YAMANA Hayato 早稲田大学, 理工学術院, 教授 (40230502)
• Co-Investigator(Renkei-kenkyūsha): OYAMA Keizo 国立情報学研究所, 教授 (90177022) ; UNO Takeaki 国立情報学研究所, 教授 (00302977)
• Research Collaborator: OKUNO Syunya ; OKUTANI Takashi ; ASAI Hiroki ; UESATO Kazuya ; TANAKA Masahiro ; SHINOHARA Shota ; ISHIYAMA Takehiro ; Wang Lan
• Project Period (FY): 2013-04-01 – 2017-03-31
• Keywords: 著者推定 / インターネットの安全性 / SNS / tweet / 専門性推定 / 信憑性

Outline of Final Research Achievements

Since various information floods on the Internet, its credibility is becoming a social problem. In this study, we researched on authorship identification technique for short messages such as SNS, targeting to identify the authorship of the messages from among 100,000 candidates. That is, if there is some documents written in advance by the author, it is possible to estimate the writer. As a result, we have established a mechanism to find a specific user out of 100,000 SNS users with accuracy of 60% if we have only 30 messages. In addition, the probability of being able to extract in the top 10 places was 74%. This is a major contribution to the fact that other research in the world is limited to about 20% accuracy for 100,000 candidates.

Free Research Field

ビッグデータ解析