Review selection based on content quality

Consumer-generated reviews have become increasingly important in decision-making processes for customers. Meanwhile, the overwhelming quantity of review data makes it extremely difficult to find useful information from it. A considerable amount of studies have attempted to address this problem by selecting reviews that might be helpful for and preferred by users. However, the performance of existing methods is far from ideal. One reason is because of lacking effective criteria to assess the quality of reviews. In this paper, we propose two novel measures, i.e. feature relevance and feature comprehensiveness, to assess the quality of reviews in terms of review content. A review selection approach is presented to select a set of reviews with high quality based on the two measures. Experiments on real-world review datasets show that our proposed method can assess the review quality effectively to improve the performance of review selection.