OpinMiner: Extracting Feature-Opinion Pairs with Dependency Grammar from Chinese Product Reviews

With the flourish of the Web, online review is become a more and more useful and important information resource for people. As a result, automatic review mining has become a hot research topic recently. Traditional review mining based on feature extracts product feature and opinion word independently, and seldom considers their association information. In this paper, we only focus on Chinese product review. We propose a method based on Chinese dependency grammar to extract feature-opinion word pairs. Specifically, we use Chinese dependency grammar to set several rules, then we make use of these rules to extract candidate feature-opinion word pairs. Finally, we filter out mismatched feature-opinion words pairs by feature ranking and Named Entity Recognition (NER) system. Experiment shows that our method in Precision is rather high.