Comprehensive Helpfulness of Online Reviews - A Time Window Based Iterative Bayesian Approach

The helpfulness of online reviews can be defined in a number of ways, including the number of votes a review receives, the length of the review, and the emotional tendency of the review. Nevertheless, one area of online helpfulness that has been overlooked is the information provided by the followers of reviews. This includes the attributes of the followers and the emotional tendency of the followers. This paper proposes a new measure of comment helpfulness, based on the proportion of negative follow-up comments, as an extension of the original index system. Given the non-uniform distribution of followers across reviews, we elected to update the original data set using the Bayesian inference method. Furthermore, this paper enhances the existing iterative Bayesian method by introducing the concept of a time window. Ultimately, the improved model is demonstrated to exhibit superior performance.