Social media is nowadays an excellent platform for gathering consumer intelligence for supporting various business intelligence applications. Among the various user-generated content on social media websites, consumer review is a critical one. However, as the volume of consumer reviews grows rapidly, it is infeasible to analyze them manually. Design of effective social media analytic techniques becomes essential. In this paper, we concentrate on proposing a review quality estimation (RQE) method. The RQE method adopts three types of indicators (namely content features, author profile features, and social network features) and five well-known supervised learning algorithms for review quality classifier induction. The Ciao public dataset is employed for empirical evaluation of the proposed RQE method. According to our empirical evaluation results, random forest algorithm has best performance. In addition, the combination of three types of features has the best classification performance.
Yang, Chin-Sheng and Kuo, Yu-Ching, "Estimation of Consumer Review Quality" (2018). PACIS 2018 Proceedings. 258.