基于异常评分行为分析的虚假评论商品识别方法

A Method for Identifying Target Products with Fake Comments Based on Abnormal Rating Behavior Analysis

在电商平台的所有评论信息中检测虚假评论时,评论领域的多样性以及虚假评论的总体稀疏性会导致识别准确率的下降。如果首先识别出包含虚假评论的商品,再对其中的评论进行针对性检测,会大大提高识别的效率和准确性。本文提出了一种基于异常评分行为分析的虚假评论商品识别方法,在对虚假评论行为分析的基础上,采用正态分布拟合和时序数据突变点检测方法,实现对虚假评论的发现。实验结果表明,该方法可以有效地识别虚假评论目标商品。

The diversity of the product field and the overall sparsity of the fake comments will lead to the recognition accuracy decline when detecting the fake comments by directly processing all the comments on the business platform. So the identification of the target product with fake comments will greatly improve the efficiency and accuracy of recognition. To this end, an abnormal rating behavior identification method is presented to identify the target products. Based on the analysis of the fake comments, using the normal distribution fitting and the abrupt point detection of the time sequence data, the discovery of the target products is realized. Experimental results show that the proposed method can effectively identify the target products with fake comments.