Effective Entity Resolution in Product Review Domain

This paper is concerned with the problem of entity resolution in the product review domain. Specifically, given many references to product features, we would like to classify references related to one feature into a group. The product feature resolution is important to product review study, such as review ranking. To solve the problem, we propose an approach which combines two types of similarity characteristics: edit distance and context similarity. Experimental results indicate that the proposed approach resolves product features effectively and improves the performance of review ranking significantly.