PULLING OUT OPINION TARGETS AND OPINION WORDS FROM REVIEWS BASED ON THE WORD ALIGNMENT MODEL AND USING TOPICAL WORD TRIGGER MODEL

Abstract

Extracting opinion targets and opinion words from on-line reviews square measure two basic tasks in opinion mining. This paper proposes an innovative approach to conjointly extract them with graph co-ranking. First, compared to previous methods that entirely used opinion relations among words, our methodology constructs a heterogeneous graph to model 2 varieties of relations, as well as linguistics relations and opinion relations. Next, a co-ranking algorithmic program is planned to estimate the confidence of every candidate, and also the candidates with higher confidence are going to be extracted as opinion targets/words. During this method, totally different relations create cooperative effects on candidates’ confidence estimation. Moreover, word preference is captured and incorporated into our co-ranking algorithmic program. In this method, our co-ranking is customized and every candidate’s confidence is merely determined by its most well-liked collocations. It helps to boost the extraction exactitude. The experimental results on 3 information sets with totally different sizes and languages show our approach achieves higher performance than progressive strategies