A Review Selection Method for Finding an Informative Subset from Online Reviews
Jin Zhang, Cong Wang, Guoqing Chen
INFORMS Journal on Computing · 2021 · Volume 33 · Issue 1
Abstract
Concerning the information overload of online reviews, this paper models a new review selection problem called the Informative Review Subset Selection problem (namely, IRSS) and demonstrates that it is NP-hard to solve and approximate. Furthermore, a novel heuristic method (namely, Combined Search-ComS) is proposed for seeking the solution to the problem and selecting a subset of reviews, which is consistent with the original review corpus in light of mutual information entropy. The proposed method is then comprehensively examined via extensive data experiments and a user study on Amazon data. Experimental results reveal the overall superiority of the proposed method in comparison with other extant methods of concern, showing that it is an effective way to select an informative subset of online reviews. The proposed method is deemed desirable and useful for online consumers and service providers.
Keywords
review selection, distribution divergence, heuristic method, feature coverage, opinion distribution
Enable JavaScript for citation tools and additional paper interactions.