Ranking Method of Object-Attribute-Evaluation Three-Tuples for Opinion Retrieval

Authors:
Masaaki Tsuchida;Hironori Mizuguchi;Dai Kusui
Affiliations:
Common Platform Software Research Laboratories, NEC Corporation., Ikoma, Japan 630-0101;Common Platform Software Research Laboratories, NEC Corporation., Ikoma, Japan 630-0101;Common Platform Software Research Laboratories, NEC Corporation., Ikoma, Japan 630-0101
Venue:
New Frontiers in Artificial Intelligence
Year:
2009

Citing 2
Cited 0

Support-Vector Networks

Machine Learning
Probabilistic latent semantic indexing

Proceedings of the 22nd annual international ACM SIGIR conference on Research and development in information retrieval

Quantified Score

Hi-index	0.02

Visualization

Abstract

In this paper, we propose a ranking method for opinion retrieval that uses a confidence model of opinion as a three-tuple of object-attribute-evaluation . The confidence model has two characteristics for calculating confidence scores. One, the model divides a three-tuple of an opinion into two pairs: object-attribute and attribute-evaluation. Two, the confidence model evaluates an opinion simultaneously using syntactic and semantic analyses. Our experiments show that our method improves the precision of the top fifty opinions in search results by ranking based on our confidence compared with random ranking.