マイクロブログの投稿時間に着目したユーザの職業推定に関する研究  [in Japanese] Research for Reasoning Occupation of Users Using Time Posted to Microblogs  [in Japanese]

Access this Article

Search this Article

Author(s)

Abstract

マイクロブログから特定の話題に対するユーザの反応を取得する技術が研究されている.マイクロブログをソーシャルセンサとして有効活用するには,ユーザごとの特性を知る必要がある.しかし,マイクロブログでは,ユーザが属性を公開していない場合が多々あるため,ユーザごとの特性を把握できない.このことから,マイクロブログのユーザ属性を推定する研究が注目されている.しかし,既存手法では,主にマイクロブログの投稿内容にのみ着目しており,リアルタイムに発信されるマイクロブログの特性を属性推定に活かせていない.そこで,本研究では,各単位時間の投稿数に基づきユーザをクラスタリングし,投稿内容,生活習慣と投稿時間帯から職業属性を推定する手法を提案する.実証実験では,投稿内容のみを使用して推定する既存手法と,時間的特徴をも考慮する本手法について比較実験を行い,本提案手法の有用性を確認した.Research is being conducted on technology to get users' reactions to specific topics in microblogs. It is necessary to know the users' characteristics in order to effectively utilize microblogs as social sensors. However, it cannot understand the users' characteristics, because user attributes are not often to the public in microblogs. For this reason, research on estimating user attributes in microblogs has been drawing attention. However, existing methods, which merely focus on the description contents in microblogs, do not take advantage of the characteristics in microblogs that transmit in real time to estimate users' attributes. This research proposes a method for classifying the users according to number of posts per unit time and estimating the occupation attributes by description contents, lifestyle and time zone of posts. Our demonstration experiments verify usability of the proposed method by comparing the existing methods of estimating merely using description contents with the proposed method of estimating using description contents and temporal characteristics.

Research is being conducted on technology to get users' reactions to specific topics in microblogs. It is necessary to know the users' characteristics in order to effectively utilize microblogs as social sensors. However, it cannot understand the users' characteristics, because user attributes are not often to the public in microblogs. For this reason, research on estimating user attributes in microblogs has been drawing attention. However, existing methods, which merely focus on the description contents in microblogs, do not take advantage of the characteristics in microblogs that transmit in real time to estimate users' attributes. This research proposes a method for classifying the users according to number of posts per unit time and estimating the occupation attributes by description contents, lifestyle and time zone of posts. Our demonstration experiments verify usability of the proposed method by comparing the existing methods of estimating merely using description contents with the proposed method of estimating using description contents and temporal characteristics.

Journal

  • 情報処理学会論文誌データベース(TOD)  

    情報処理学会論文誌データベース(TOD) 6(5), 71-84, 2013-12-27 

    Information Processing Society of Japan (IPSJ)

Codes

  • NII Article ID (NAID)
    110009656663
  • NII NACSIS-CAT ID (NCID)
    AA11464847
  • Text Lang
    JPN
  • Article Type
    Article
  • ISSN
    1882-7799
  • Data Source
    IPSJ 
Page Top