短句语义向量计算方法  

Vector semantic computing method study for short sentence

作　　者：陈福[1] 林闯[2] 薛超[2] 徐月梅[1] 孟坤[2] 倪艺函 

机构地区：[1]北京外国语大学计算机系,北京100089 [2]清华大学计算机系,北京100084

出　　处：《通信学报》2016年第2期11-19,共9页Journal on Communications

基　　金：国家自然科学基金资助项目(No.61170209;No.61173008;No.61502038;No.61370132);教育部新世纪优秀人才支持计划基金资助项目(No.NCET-13-0676);2011重点课题基金资助项目(No.BFSU2011-ZS04)~~

摘　　要：提出了一种基于人工神经网络的短文语义向量放缩算法,结合社交节点自身信息和短文语义,给出社交网络短文语义计算方法和突发话题发现算法。通过文本数值化实现语义距离的计算、比较、节点的分类及社区发现等。通过自行开发的微博采集工具Argus采集的大量新浪微博内容对所提模型和算法进行了验证,最后对未来工作进行了展望。A vector semantic computing method study for short sentence based on artificial neural network was proposed. And a semantic computational algorithm for social network texts as well as a discovery algorithm for emergencies was provided with reference to the information provided by the social nodes itself and the semantic of the text. Through the numerization of text, the calculation and comparison of semantic distance, the classification of nodes and the discovery of community can be realized. Then, huge quantities of Sina Weibo contents are collected to verify the model and algorithm put forward. In the end, outlooks for future jobs are provided.

关 键 词：在线社会网络 主题语义计算 人工神经网络 突发话题发现 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]