一种基于自注意力机制的用户评论情感分类方法

A User Comment Sentiment Classification Method Based on Self Attention Mechanism

近年来,用户评论情感分类方法成为自然语言处理领域的重要研究内容。本文利用自注意力机制在文本中捕捉重要局部特征的优势,在没有外部语法信息的条件下自动学习上下文关系,并结合卷积神经网络模型TextCNN,提出一种基于自注意力机制的文本分类模型(TextCNN Attention,TextCNN-Att),TextCNN-Att模型结合了卷积神经网络中的卷积操作,但不对卷积结果进行Maxpoolling,而是对卷积结果进行Attention处理,然后对每一个Attention结果进行Softmax,从而实现对用户评论的情感分类。实验结果表明,TextCNN-Att在2018 AI全球挑战赛的数据集上的精确率为0.921,召回率为0.945,F1值为0.933。

In recent years,the sentiment classification method of user comments becomes an important research content in the field of natural language processing.This paper takes advantage of the self-attention mechanism to capture important local features in the text,automatically learns the contextual relationship without external grammatical information,and combines the text classification algorithm TextCNN to propose a text classification algorithm based on the self-attention mechanism(TextCNN Attention,TextCNN-Att),the TextCNN-Att algorithm combines the convolution operation in the convolutional neural network,but does not perform maxpoolling on the convolution results,but processes the convolution results with attention processing,and then performs softmax on each attention result,so as to realize the sentiment classification of user comments.The experimental results show that TextCNN-Att has an accuracy rate of 0.921,a recall rate of 0.945 and an F1 value of 0.933 on the 2018 AI Global Challenge data set.