Study On False Comments And Identification Of E-Commerce Platform Shopping

With the development of the Internet and the consumer demand for diversification of consumer channels,the e-commerce platform has developed rapidly and online shopping has increasingly become a mainstream way for consumers to purchase.Consumers use commodity descriptions and prices to examine product quality and other information through product review information.Some merchants in order to attract customers,increase customer's browsing,attention and purchase,paid to hire a group of network to brush one hand,so that they can create some good reviews without promoting and using the actual goods in order to promote consumer favorability of the product Promote to buy goods.This will undoubtedly affect the consumer's decision and harm consumers' rights and interests.Therefore,the accurate identification of false comments is an important and urgent task.The purpose of this paper is to give a set of methods and processes for accurately identifying false comments,and to investigate the mode of false comments.The data mining method is used to identify false comments.The main tasks include: acquiring sample product data from different e-commerce platforms,quantifying the text,and pre-identifying false comments through information such as comment time,repeated comments,and reviewer rating;Logistic regression,k-nearest neighbor model,SVM,text-CNN,fast Text,and combined model are used to accurately identify false comments;then,through a large amount of data,the false commentary model is examined to construct a language model for false comments,and multidimensional features are used.Investigate the behavioral properties of false comments to unearth the pattern of false comments on behavioral attributes.The innovations in this paper include: 1.Precognition of false comments through multidimensional features of data such as repeated comments and commentary time distribution,manual annotation and subsequent analysis combined with pre-identification results;2.Adjustment of traditional model algorithms.In addition,the model's classification effect gives weight to the model,integrates the model,and enhances the effect of false comment recognition.3 In addition to this,the paper establishes a language model for false comment information through false comment recognition results and analyzes the multi-dimensional features of false comments.To examine the pattern of behavior of false comments.