Study On Identification Method Of The Fake Reviewers In C2C E-commence

The growing fake reviews seriously hamper the development of the domestic e-commerce, which also mislead customers to make wrong decision, and even bring loss in economy, so a detecting method of fake reviewers is needed. Based on user characteristics, this thesis proposes the detecting method of fake reviewers which can effectively detect fake reviewers and bring customers a sound credit environment.Based on the method of vector representation, this thesis studies the user’s characteristic model of C2C e-commerce, analyzes the different concerns of buyers and sellers in the process of spam reviewing, and builds the buyer’s characteristic model based on buyer’s credit rating, reviewing time, control degree of review, etc. and the seller’s characteristic model based on seller’s credit rating, praised degree, etc. To lay the foundation for fake reviewer detecting, the different features between spam reviewers and non-spam reviewers, as well as the different between dishonest sellers and honest sellers are analyzed and quantified.Based on user characteristics, this thesis then proposes a fake reviewer detecting method. Firstly, for the sake of higher efficiency and accuracy, it proposes a method for exception analysis of fake reviewer detection based on the differences of multiple features between sellers. Secondly, according to the similarity between buyers, a buyer’ similarity network can be built. After analyzing cliques on the network, a set of candidate groups including spammers and non-spammers can be gotten. Finally, spammer groups can be found according to co-review number between spammers and non-spammers. On this basis, fake reviewers can be identified easily. The experimental results show that the fake reviewer detecting method based on user characteristics is feasible and effective.