Opinion Feature Extraction Using Class Sequential Rules

Minqing Hu, Bing Liu

The paper studies the problem of analyzing user comments and reviews of products sold online. Analyzing such reviews and producing a summary of them is very useful to both potential customers and product manufacturers. By analyzing reviews, we mean to extract features of products (also called opinion features) that have been commented by reviewers and deter-mine whether the opinions are positive or negative. This paper focuses on extracting opinion features from Pros and Cons, which typically consist of short phrases or incomplete sen-tences. We propose a language pattern based approach for this purpose. The language patterns are generated from Class Se-quential Rules (CSR). A CSR is different from a classic se-quential pattern because a CSR has a fixed class (or target). We propose an algorithm to mine CSR from a set of labeled training sequences. To perform extraction, the mined CSRs are transformed into language patterns, which are used to match Pros and Cons to extract opinion features. Experimental results show that the proposed approach is very effective.


This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.