DETECTING AND CHARECTERIZING EXTREMIST REVIEWER - GROUPS IN ONLINE

Online reviews have come an important source of information for druggies before making an informed purchase decision. Beforehand reviews of a product tend to have a high impact on the posterior product deals. In this paper, we studied the characteristics of early detecting through their posted reviews on two real- world large-commerce platforms, i.e., Amazon and Yelp. In specific, we divide product continuance into three successive stages, videlicet beforehand, maturity and dalliers. A stoner who has posted a review in the early stage is considered as an early critic. We quantitatively characterize early pundits grounded on their standing actions, the helpfulness scores entered from others and the correlation of their reviews. We have plant that (1) an early critic tends to assign a advanced average standing score; and (2) an early critic tends to post further helpful reviews. Our analysis of product reviews also indicates that early pundits’ conditions and their entered helpfulness scores are likely to impact of product. By viewing review posting process a multiplayer competition game, we propose a new periphery- grounded embedding model for early critic vaticination. Expansive trials on two different-commerce datasets have shown that our proposed approach outperforms a number of competitive nascence.