基于多维特征权重的虚假评论识别方法

Recognition of Fake Evaluation Based on Multi-dimensionality Feature Weight

在线商品评论是消费者网购决策的重要依据,利益的驱动使得越来越多的网络虚假评论呈现在消费者面前。针对此问题,提出一种多维特征权重的在线虚假评论识别方法。首先,从网购信息有用性角度出发,在商品、评论者和评论内容3个维度中选取9个对评论属类语义贡献大的特征。然后,基于Fisher准则,运用赋予权重的特征构建用于识别虚假评论的方法。试验结果验证了基于多维特征权重的虚假评论识别方法的有效性:多维特征权重方法的准确率。

Online goods evaluation is important basis for consumers to make a purchase decision.Consequently,consumers faces with more and more fake evaluations.In consideration of the problem,an online fake evaluation based on multi-dimensionality feature weighting is presented.Firstly,in this paper the usefulness of information from online shopping is analyzed,and nine important features that are conducive for the classification of three dimensions of commodity,reviewer and reviews are chosen.Secondly,the feature weighting based on Fisher Criterion is used for detecting fake evaluation.Finally,the experimental results show that the propose method is effectively.