摘要
为了有效地提取评论文本特征,进行虚假信息的检测,采用卷积神经网络的方法进行虚假评论的识别。文章基于扩展Ott黄金数据集,通过word2vec将评论语料转换为词向量作为CNN的输入;按照虚假评论检测的实验效果,确定了卷积神经网络的向量维度和网络深度结构,形成卷积神经网络的优化模型。在同一数据集上与LSTM和GRU算法模型进行了对比实验,结果表明,卷积神经网络在虚假评论检测中有效。
In order to extract the features of comment text effectively and detect false information, this paper uses the method of convolutional neural network to recognize false comment. With the extended Ott gold data set, the comment corpus is converted into the word vector by word2 vec as the input of CNN. According to the experimental results of false comment detection, the vector dimension and network depth structure of convolution neural network are determined to form a optimized model of convolution neural network. A comparative experiment on the same data set is carried on with LSTM and GRU algorithm models,the results show that the convolutional neural network is effective in false comment detection.
作者
黄欣欣
年梅
胡创业
范祖奎
Huang Xin xin;Nian Mei;Hu Chuangye;Fan Zukui(College of Computer Science and Technology, Xinjiang Normal University, Urumqi, Xinjiang 830054, China;Department of Language, Xinjiang Police College)
出处
《计算机时代》
2019年第11期41-45,共5页
Computer Era
基金
赛尔网络下一代互联网技术创新项目（NGII20160604）
新疆师范大学重点学科资助项目（17SDKD1201）
国家自然科学基金（61771089）
