基于双卷积神经网络的虚假评论识别

Spam Review Detection Based on Double Convolutional Neural Network

传统的虚假评论识别方法大多采用机器学习算法,并把虚假评论识别当作一个二分类的任务进行处理,对数据集中的文本数据提取一些关键的特征,并使用机器学习的算法对提取的特征进行训练,从而达到分类的目的。在研究虚假评论特点的基础上,使用卷积神经网络分别对评论文本数据以及评论者行为数据进行处理,融合了评论文本信息和评论者行为信息,提出了基于双卷积神经网络的虚假评论识别方法,经实验验证该方法对虚假评论的识别有较高的准确率。

Most of the traditional false comment recognition methods use machine learning algorithms,and treat false com⁃ment recognition as a binary classification task,extract some key features from the text data in the data set,and use machine learn⁃ing algorithms to perform the extracted features to train the extracted features to achieve the purpose of classification.On the basis of studying the characteristics of false reviews,the convolutional neural network is used to process the review text data and reviewer be⁃havior data separately,and the review text information and reviewer behavior information are combined,and a false review recogni⁃tion based on double convolutional neural network is proposed.Method,it is verified by experiments that this method has a high ac⁃curacy rate in identifying false reviews.