| With the popularization and rapid development of Internet technology,e-commerce platforms are more widely used in real life,such as Taobao,Jingdong,Xiaohongshu,etc.As a result,online reviews have become more and more popular as a social media.The main manifestation is that when consumers buy goods,they will regard the historical review records of the goods as an important reference index.Every seller strives to improve the praise of their products to attract more consumers.Driven by interests,some merchants may employ some people to make false comments on their products,which will be used to improve the popularity of the shops and the favorable comments of the products.Some merchants may employ some people to maliciously suppress the products of their peers to improve their competitiveness.In order to provide a good shopping environment for every consumer,false comments detection has gradually become an important link in maintaining the normal operation of e-commerce platforms and online shopping websites.This paper proposes an online detection method for false comments based on Markov random fields.Compared with previous offline detection methods,online detection has higher real-time performance and can capture false comments in a more timely manner,thus being applied to actual scenes more efficiently.The content of this experiment mainly includes: preprocessing the acquired data set,the main purpose of which is to facilitate the data searching in the experiment process and the data processing in the encoding process;use the relationship between reviews to generate a dynamic review graph,and then model it as a Markov random field;use LBP algorithm to solve the Markov random field.The experimental results show that the online detection method of false comments based on Markov random field proposed in this paper has a high accuracy and recall rate after parameter tuning,which surpasses most of the existing methods to solve similar problems. |
PDF Full Text Request