基于双通道卷积神经网络的文本情感分类算法

Text Sentiment Classification Algorithm Based on Double Channel Convolutional Neural Network

针对现有深度学习方法在文本情感分类任务中特征提取能力方面的不足,提出基于扩展特征和动态池化的双通道卷积神经网络的文本情感分类算法.首先,结合情感词、词性、程度副词、否定词和标点符号等多种影响文本情感倾向的词语特征,形成一个扩展文本特征.然后,把词向量特征与扩展文本特征分别作为卷积神经网络的两个输入通道,采用动态k-max池化策略,提升模型提取特征的能力.在多个标准英文数据集上的文本情感分类实验表明,文中算法的分类性能不仅高于单通道卷积神经网络算法,而且相比一些代表性算法也具有一定的优势.

The existing deep learning method is insufficient to extract features in the text sentiment classification task. To solve the drawback, a text sentiment classification algorithm based on the double channel convolutional neural network with extended features and a dynamic pooling is presented. Firstly, various word features influencing the sentiment orientation of text, such as emotional word, part of speech, adverb of degree, negative word and punctuation, are combined to obtain an extended text feature. Then, the word vector feature and the extended text feature are used as two individual channels of the convolutional neural network, and a new dynamic k-max pooling strategy is adopted to improve the efficiency of feature extraction. The experimental results on standard English datasets demonstrate that the proposed algorithm achieves better classification efficiency than traditional convolutional neural network algorithm with single channel, and it is more advantageous compared with some elitist text sentiment classification algorithms.