A novel Chinese text mining method for E-commerce review spam detection

Abstract

Review spam is increasingly rampant in China, which seriously hampers the development of the vigorous e-commerce market. In this paper, we propose a novel Chinese text mining method to detect review spam automatically and efficiently. We correctly extract keywords in complicated review text and conduct fine-grained analysis to recognize the semantic orientation. We study the spammers’ behavior patterns and come up with four effective features to describe untruthful comments. We train classifier to classify reviews into spam or non-spam. Experiments are conducted to demonstrate the excellent performance of our algorithm.