摘要
针对传统文本分类过程中词表示特征时不够全面、可解释性差的问题,提出一种基于词和事件主题的W-E CNN文本分类方法,并给出基于BTM的事件主题模型。将传统基于词的特征表示方法与事件主题特征表示方法进行拼接作为CNN的输入,丰富特征语义信息,提高了文本分类的准确性。实验分析可知,该方法的分类准确性在一定程度上要优于其他方法。
In order to solve the problem that the traditional text classification process is not comprehensive and interpretable,this paper proposes a W-E CNN text classification method based on words-event topic,and gives an event topic model based on BTM.The traditional word-based feature representation method and event topic feature representation method were spliced as the input of CNN,which enriched the feature semantic information and improved the accuracy of text classification.The experimental analysis shows that the classification accuracy of this method is better than other methods to a certain extent.
作者
于游
付钰
吴晓平
Yu You;Fu Yu;Wu Xiaoping(Department of Information Security,Naval University of Engineering,Wuhan 430033,Hubei,China)
出处
《计算机应用与软件》
北大核心
2021年第5期170-174,240,共6页
Computer Applications and Software
基金
国家自然科学基金项目(61672531)。
