摘要
短文本具有长度短、特征稀疏以及上下文依赖强等特点,传统方法对其直接进行分类精度有限。针对此问题,提出了一种结合字符和词的双输入卷积神经网络模型CP-CNN。该模型通过加入一种用拼音序列表征字符级输入的方法,构建字符级和词级的双输入矩阵,并在采样层使用k-max采样方法,增强模型特征的表达能力。利用豆瓣电影评论数据集对该模型进行识别精度评估,实验结果表明,与传统分类模型和标准卷积神经网络模型相比,该模型可有效提高短文本分类效果。
Since short text is characterized of the short length,sparse features and strong context dependency,the traditional models have a limited precision.Motivated by this,this paper proposed a multi-input convolutional neural network model CP-CNN.It used pinyin sequences to characterize the feature at the character level,thus to build double input matrix at the character and phrase level.It could enhance the model’s feature presentation ability by using k-max down-sampling method.The evaluations on Douban review dataset show that the proposed model outperforms the standard CNN and traditional models on short text classification.
作者
余本功
张连彬
Yu Bengong;Zhang Lianbin(School of Management,Hefei University of Technology,Hefei 230009,China;Key Laboratory of Process Optimization&Intelligent Decision-making of Ministry of Education,Hefei University of Technology,Hefei 230009,China)
出处
《计算机应用研究》
CSCD
北大核心
2018年第4期1001-1004,共4页
Application Research of Computers
基金
国家教育部人文社会科学基金资助项目(2012JYRW0710)
国家自然科学基金资助项目(71671057)
关键词
短文本
分类
卷积神经网络
short text
classification
convolutional neural network(CNN)
