In this paper, we propose a object retrieval method to deal with review's reliability search for reviewed object. We believe that using reviews is effective to search reviewed objects, because reviews are considered as evaluation of reviewed objects. But, reviews are not always reliable. Therefore considering review reliability is needed. In this paper, we calculate review's support level using reviewer's votes that submitted by the other users. We focused a case that review's support level are the same because reliability of support level is different by the number of votes. We propose a calcuating method of review's reliability by applying sampling error. Finally, we compare proposal method with the method of TFIDF by 11-point average precision.
[1]
David M. Pennock,et al.
Mining the peanut gallery: opinion extraction and semantic classification of product reviews
,
2003,
WWW '03.
[2]
Oren Etzioni,et al.
Extracting Product Features and Opinions from Reviews
,
2005,
HLT.
[3]
Peter D. Turney.
Thumbs Up or Thumbs Down? Semantic Orientation Applied to Unsupervised Classification of Reviews
,
2002,
ACL.
[4]
Satoshi Morinaga,et al.
Mining product reputations on the Web
,
2002,
KDD.
[5]
Gerard Salton,et al.
Automatic Text Processing: The Transformation, Analysis, and Retrieval of Information by Computer
,
1989
.