Research on Fake Rating Detection Algorithm Based on Deep Learning

In the e-commerce market, fake rating detection has been one of the key and difficult problems that disturb the e-commerce platforms. There have been a large number of researches on this topic all over the world, in which many algorithms and models have been proposed. Most of these algorithms are supervised and semi-supervised. However, few e-commerce platforms are willing to disclose their own transaction data and detection methods, thus it is challenging to perform theoretical studies on the detection algorithms. In order to address this challenge, this paper proposes a semi-supervised fake rating detection method based on deep learning (FRDDL) and e-commerce platform reputation evaluation method under the premise of using a small number of training samples and only considering the behavior characteristics of the buyers. The experimental results on the hotel rating data in Yelp showed that the proposed algorithm had better detection performance than the existing algorithms.