摘要
针对在自然语言处理中起着关键作用的文本相似度计算问题,提出了一种神经网络深度学习的词向量模型计算方法.利用词向量计算文本语义相似度,并采用高频词滤波的方法削弱扰动的影响.对百度新闻、新浪新闻等的中文词库进行训练,并与传统的检测方法进行对比.实验结果证明了提出方法的有效性和准确性.
A novel method based on deep-learning of neural work is proposed to deal with the text similarity computing in natural language processing.The word vector model is used to calculate the mantic similarity,and the high-frequency word filtering-based method is adopted to reduce the interference.By training the Chinese corpus from Baidu News and Sina News,and comparing with the traditional method,the experiment results show the effectiveness and accuracy of the proposed method.
作者
汪一百
陈实
叶剑锋
WANG Yi- bai;CHEN Shi;YE Jian- feng(Key Laboratory Breeding Base of Hunan Oriented Fundamental and Applied Research of.Innovative Phannaceutics,Changsha 410219;School of Information Engineering,Changsha Medical University,Changsha 410219;School of Mechanical and Electrical Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 210016 China)
出处
《湘潭大学自然科学学报》
CAS
2018年第2期104-107,共4页
Natural Science Journal of Xiangtan University
基金
湖南省教育厅科学研究项目(16C0184)
新型药物制剂研发湖南省重点实验室培育基地(2016TP1029)
关键词
文本相似度计算
词向量模型
深度学习
高频词滤波
text similarity computing
word vector model
deep learning
high-frequency word filtering
