Research On Recognition Of Spam Reviews On E-commerce Platforms Based On GCN

With the rapid development of the e-commerce industry,more and more consumers choose to buy products and services on the e-commerce platform,and the evaluation of the product is an important reference for consumers to choose whether to purchase.Fake reviews are hidden in the massive review data of e-commerce platforms,which can cause serious interference to consumers' effective choices of products.In order to further realize the effective identification of false reviews,the comment data of the e-commerce platform is used as the entry point to construct annotated Chinese false reviews data set,based on a two-layer graph convolutional network to establish a false review recognition model,and Based on the model,an e-commerce platform false comment recognition system was designed and implemented.In order to solve the previous problem of the lack of Chinese annotation data set for false comment recognition,a Chinese false comment data set was constructed.Use web crawlers to automatically collect a large number of product review data on the JD platform.After data filtering,word segmentation,and stop word filtering are performed on the review text in the dataset,the machine learning classifier combined with Active Learning(AL)is comprehensively applied And repeated comment recognition to complete the annotation of the comment data,effectively constructing the data set.Using Graph Convolutional Networks(GCN,Graph Convolutional Networks)to establish a false comment recognition model.Use the comment text to construct the comment-word heterogeneous graph as the model input,and send it to the two-layer GCN network for training.After optimizing the model,a false comment recognition experiment was carried out using the constructed data set.Compared with Text CNN and Text RNN,the model can obtain a higher accuracy rate,which verifies the effectiveness of the model.Based on the false comment recognition model,an e-commerce platform false comment recognition system is designed and implemented.The user submits the comment text in the system,and the system will distinguish and display the authenticity of the comment.