双通道卷积记忆神经网络文本情感分析

近年来，电子商务发展迅速，对电商商品评论进行情感分析可为消费者购物、商家调整销售策略与电商平台个性化推荐提供重要参考意见，因此提出双通道卷积记忆神经网络文本情感分析模型。首先，通过词向量与由特征词典构造的扩展特征矩阵两个不同的通道进行卷积运算，再利用卷积神经网络提取文本局部最优信息，最后利用长短期记忆神经网络学习长距离的上下文情感，完成文本情感分析任务。实验结果表明，与多种文本情感分析方法相比，双通道卷积记忆神经网络文本分析算法具有较高的精度，达到95%，且考虑了文本语义信息与文本情感信息，可获得更好的文本表示，同时兼顾文本局部特征与上下文信息的学习，可有效提高文本情感分析准确率。

Text Sentiment Analysis of Dual-Channel Convolutional Memory Neural Network

In recent years， e-commerce has developed rapidly. The purpose of sentiment analysis of e-commerce reviews is to provide an important reference for consumers to buy， merchants to adjust sales strategy and personalized recommendation of e-commerce platform. A text sentiment analysis algorithm based on the double channel convolution memory neural network is presented. Firstly， the convolution operation was carried out by using two different channels， namely， the word vector and the extended feature matrix constructed by feature dictionaries. Secondly， convolution neural network was used to extract the local optimal information of the text. Finally， long-term and short-term memory neural network was used to learn long distance context sentiment so as to complete text sentiment analysis task. Experimental results show that the proposed algorithm has higher accuracy compared with many text sentiment analysis methods. Dual-channel convolutional memory neural network text analysis algorithm considers the semantic information and emotional information of text to get better text representation. It also takes into account the local features of the text and the learning of context information， which can effectively improve the accuracy of text emotional analysis.