著者

Takama Yasufumi Mao Zhongjie Hattori Shunichi

出版者

富士技術出版株式会社

雑誌

Journal of Advanced Computational Intelligence and Intelligent Informatics (ISSN:13430130)

巻号頁・発行日

vol.18, no.3, pp.331-339, 2014

---

<p>This paper proposes a method for classifying informative reviews based on personal values. Reviews of an item are useful for a user who is considering purchasing it. However, it is difficult for readers to find informative reviews from vast amount of reviews because of existence of too many uninformative reviews. This paper supposes that the value of a review is affected by reader-dependent and independent factors. Typical uninformative reviews in terms of reader-independent factor are copy-and-paste reviews, which do not provide any readers with useful information for their decision-making. On the other hand, it is supposed different readers regard different reviews as informative, which is affected by their personal values. This paper focuses on such a reader-dependent factor, and proposes a methods for classifying informative reviews based on reader's personal value. Experiments are conducted using actual review data provided by Rakuten Inc., of which the results show about 0.7 of average accuracy is achieved. Furthermore, it is also shown proposed method can model judging criteria common to those who have similar personal values.</p>

著者

桑折 章吾 加藤 優 高間 康史 Kori Shogo Kato Yu Takama Yasufumi

雑誌

SIG-AM = SIG-AM

巻号頁・発行日

vol.04, no.02, pp.9-14, 2013-07-29

---

本稿では,検索エンジンを用いた情報検索におけるユーザ行動を分析した結果について報告する.我々は,「動向に関する問い」を対象とした検索エンジン構築を目指し,その基本的検索機能について検討を進めている.既存検索エンジンを用いた検索でも,ユーザは異なる意図に基づく基本的検索を組み合わせて目的を達成しているとの考えに基づき,本稿では検索意図の観点からユーザの情報検索行動を分析し,得られた結果に基づき動向に関する基本的検索機能について考察する.