ブログにおける情緒対象の分析システムの試作  [in Japanese] A Prototype System to Analyze Emotional Targets in Blogs  [in Japanese]

Search this Article

Author(s)

    • 徳久 雅人 TOKUHISA Masato
    • 鳥取大学大学院 工学研究科 情報エレクトロニクス専攻 Department of Information and Electronics, Graduate School of Engineering, Tottori University

Abstract

本稿では,人々が感じている情緒の対象をブログから分析するシステム(ABLANA)について述べる.本システムは,ブログサイトから発信されるRSSに基づきブログを収集し,パターンを用いた情緒推定により,情緒名とともに情緒対象を抽出する.パターン(約14,800件)は,情緒原因,情緒状態,および,情緒表出を解析することができる.日本語のブログサイト3社より約半年ぶんのブログ記事(約1億文)を収集し,分析実験を行った.RSSを用いることで,ブログ記事を時系列で収集することができた.情緒対象の出現頻度についてのZ-スコア,および,情緒対象と情緒の共起確率を用いることで,人々の関心や不満の生じた時期を特定することができた.

This paper discusses a prototype system of an affective blog analyzer (ABLANA), which analyzes emotional states and extracts emotional targets of people from blogs. ABLANA uses RSS sent by blog sites to collect blog articles and then estimates emotions and their targets from the articles by a pattern dictionary. The dictionary contains approximately 14,800 patterns which can analyze the emotional cause, state and responsive expression. In the experiments, blog articles were collected from three Japanese blog sites during six months, which contained approximately 100,000,000 sentences. As the result, blog articles were ordered by the written time by using RSS. The aroused term of interest and dissatisfaction were successfully clarified by using both the Z-score of the appearance of keyword and the probability of the co-occurrence of the triplet of emotion-target-term.

Journal

  • IEICE technical report

    IEICE technical report 109(253), 1-6, 2009-10-17

    The Institute of Electronics, Information and Communication Engineers

References:  24

Cited by:  3

Codes

  • NII Article ID (NAID)
    110007483538
  • NII NACSIS-CAT ID (NCID)
    AN10449078
  • Text Lang
    JPN
  • Article Type
    Journal Article
  • ISSN
    09135685
  • NDL Article ID
    10434004
  • NDL Source Classification
    ZN33(科学技術--電気工学・電気機械工業--電子工学・電気通信)
  • NDL Call No.
    Z16-940
  • Data Source
    CJP  CJPref  NDL  NII-ELS 
Page Top