Extraction of Opposite Sentiments in Classified Free Format Text Reviews

Opinion mining received well attention in finding personal opinions from user generated content. These opinions contain valuable information for improving and/or comparing the products or services. On the other hand, given a customer review that the opinion has been already classified into a certain sentiment polarity, for instance positive or negative, the opposite sentiments are more and more interesting for the decision makers. In this paper, we propose an unexpected sequence mining based approach to extract opposite sentiments from classified free format text reviews. We first adapt the notions in sequence mining to the opinion mining, then we represent the sentence-level sentiments as the sequential implication rules and from which we generate the belief system for formalizing opposite sentiments. We further propose the algorithm MOSUS (Mining Opposite Sentiments as Unexpected Sentences) that extracts opposite sentiments with respect to the belief system. We conclude by detailing ...

• Type: Conference
• Year: 2008
• Where: DEXA
• Authors: Dong (Haoyuan) Li, Anne Laurent, Mathieu Roche, Pascal Poncelet
• Keywords: Database, DEXA 2008, Mining Opposite Sentiments, Opinion Mining, Opposite Sentiments