Identifying the Features of Reputable Users in eWOM Communities by Using Particle Swarm Optimization

Electronic Word-of-Mouth communities have become popular over the last several years as websites where people can share their online reviews about any type of product or service. As a mechanism to improve trust, posted reviews can also be scored by the rest of the community in terms of helpfulness, so users can reach a high level of reputation through their interactions with other users and can thus increase their credibility. The aim of this paper is to investigate the main patterns of activity that characterize reputable users by using a set of classification rules. However, the class of reputable users is only a fraction of the total number of users. Due to the imbalance between the classes, i.e., reputable and non-reputable users, and the high dimensionality of the problem, an evolutionary computation algorithm such as Particle Swarm Optimization (PSO) is applied to obtain the main activity patterns of reputable users. Obtained results can help us better understand the mechanism of trust in eWOM communities and avoid the undesirable manipulation of reputations by false accounts.