Research On The Method For Recognizing Fake Reviews In The Automobile Forum Based On Demand Analysis

The forum is a significant channel used to facilitate the communication among users and the reviews in the forum have become a important information resource for enterprises to discover preferences of the public and market trend.However,the forum also is the hidden site for enterprise marketing and malicious attacking against competitors.Therefore,finding out the spammers in the forum has become the significant and complex issue,and it will be practical and valuable when we discover the forum users' demand information based on filtering out fake reviews in the forum.At present,the number of fake reviews in the forum is growing rapidly and it has become the biggest obstacle for forum users to refer to the information and enterprises acquire the real feedback of forum users.In order to recognize forum spammers more accurately,the paper pioneers to classify the forum spammers into automated forum spammers and marketing forum spammers based on different behavioral features among forum spammers.Besides,the paper constructs five-step cascading recognition model based on the statistical method: Recognition Automated Spammer(RAS)and recognition model based on the k-Means clustering method: Recognition Marketing Spammer(RMS).Then,the paper constructs the text mining-driven product demand information recognition model to discover demand information in the user reviews.In order to demonstrate the effectiveness of models the paper constructed,the paper uses automobile forum data to test the models.The final experimental result also demonstrates that the behavior-driven forum spammer recognition model has the high accurate rate and recall rate.At the same time,the product demand information recognition model with sentiment analysis and information gain methods also recognizes notable product features that are influential for user satisfaction in the user reviews.Besides,to our best knowledge,this study is among the first to construct behavior-driven recognition models according to the different behavioral patterns of the forum spammers respectively,it has the significant value in the filed that recognizing forum spammers.