基于多特征改进卷积神经网络的文本情感分析  

Sentiment Analysis Model Based on Multi-feature Improved Convolutional Neural Network

作　　者：付晓杰 张曦煌[1] FU Xiaojie;ZHANG Xihuang(School of Internet of Things Engineering,Jiangnan University,Wuxi 214000)

机构地区：[1]江南大学物联网工程学院,无锡214000

出　　处：《计算机与数字工程》2021年第3期536-541,共6页Computer & Digital Engineering

基　　金：江苏省产业学研究合作项目基金(编号:BY2015019-30)资助。

摘　　要：目前基于端到端的卷积神经网络在文本情感分析中大量应用,但是其缺乏对罕见词的处理以及对不同领域的泛化性。基于此提出了一种加入词典特征以及词性特征进行特征增强的卷积神经网络模型——多特征改进卷积神经网络模型。首先采用字特征、词典特征以及词性特征表示句子向量,其次采用分段多池操作抽取重要特征,最后采用投票机制判段文本的情感倾向性。实验中,在数据集相同的情况下与多种情感分析网络模型的进行对比,MFICNN模型的情感识别精度最高,达到了0.944。At present,end-to-end convolutional neural networks are widely used in text sentiment analysis,but it lacks the processing of rare words and the generalization of different fields.Based on this,a multi-feature improved convolutional neural net⁃work(MFICNN)model is proposed,which is based on the feature of the dictionary and the feature of the part of speech for feature enhancement.Firstly,the character feature,dictionary feature and part-of-speech feature are used to represent the sentence vector.Secondly,the segmented multi-pool operation is used to extract the important features.Finally,the voting mechanism is used to judge the sentiment orientation of the text.In the experiment,compared with the various sentiment analysis network models under the same dataset,the MFICNN model has achieved the highest emotion recognition accuracy,reaching 0.944.

关 键 词：卷积神经网络 文本情感分析 分段多池 投票机制 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]