Extracting opinion targets and opinion words from online reviews with graph co-ranking

Abstract

Extracting opinion targets and opinion words from online reviews are two fundamental tasks in opinion mining. This paper proposes a novel approach to collectively extract them with graph coranking. First, compared to previous methods which solely employed opinion relations among words, our method constructs a heterogeneous graph to model two types of relations, including semantic relations and opinion relations. Next, a co-ranking algorithm is proposed to estimate the confidence of each candidate, and the candidates with higher confidence will be extracted as opinion targets/words. In this way, different relations make cooperative effects on candidates' confidence estimation. Moreover, word preference is captured and incorporated into our coranking algorithm. In this way, our coranking is personalized and each candidate's confidence is only determined by its preferred collocations. It helps to improve the extraction precision. The experimental results on three data sets with different sizes and languages show that our approach achieves better performance than state-of-the-art methods. © 2014 Association for Computational Linguistics.