卷积神经网络在短文本情感多分类标注应用  

Application of convolutional neural network in multi-category classification for short text sentiment

作　　者：周锦峰 叶施仁[1] 王晖[1] ZHOU Jinfeng;YE Shiren;WANG Hui(School of Information Science and Engineering,Changzhou University,Changzhou,Jiangsu 213164,China)

机构地区：[1]常州大学信息科学与工程学院,江苏常州213164

出　　处：《计算机工程与应用》2018年第22期133-138,149,共7页Computer Engineering and Applications

基　　金：国家自然科学基金(No.61272367);江苏省科技厅项目(No.BY2015027-12)

摘　　要：情感多分类标注对文本信息的敏感性远高于二分类问题。为了有效利用语义依赖距离和语义多层次进行情感多分类,提出一种多窗口多池化层的卷积神经网络模型。首先使用多窗口的卷积层提取上下文局部语义,然后通过多池化层降低特征维度,同时保留不同层次的语义,由多层次语义构成文本特征向量,最后送入全连接层完成多分类标注。采用斯坦福情感树库数据集验证所提模型的多分类标注效果。实验结果表明,在训练集含短语和未包含短语两种设定下,模型的短文本情感多分类正确率分别达到54.6%和43.5%。Deep learning based approaches achieved less for sentiment classification with multiple labels.For this issue,this paper proposes a model called mwmpCNN(multi-windows and multi-pooling Convolutional Neural Network)to grasp the semantic distance and various emotional levels.mwmpCNN assemblies convolution layer with multiple windows to extract local context semantic,and then applies multi-pooling layer to keep multi-level semantic in short text when the feature dimension is reduced.Here,the text feature vector is constructed and reflected by the multi-level semantic,and connection layer is implemented for multi-label classification.This paper evaluates mwmpCNN by the test on Stanford Sentiment Treebank.mwmpCNN exhibits the classification accuracy of 54.6%and 43.5%respectively for the multilabel classification task.

关 键 词：情感分析 多分类标注 卷积神经网络 深度学习 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]