摘要
微博是社交网络的主要形式,其短文本和时效性的特点能够体现出当前大众最新兴趣取向.微博文本不同于传统文本,其时效性的特点使得在对其进行主题挖掘时容易忽略时间因素而造成结果不准确.针对此问题,提出了采用可变时间窗口的TIF-LDA微博主题模型对微博主题分析做出时间限定,并基于微博发布的时间为微博词条添加时间权重,使用词条的时间权重之和作为词条在LDA主题挖掘计算中的影响因子.实验结果表明,相较于标准的LDA主题模型,本文所提模型能够更加准确地反映用户最新的关注热点.
Micro-blog is the main form of social network,and its short text and timeliness can reflect the latest interest orientation of current public.The text of Micro-blog is different from the traditional text,its timeliness makes it easy to ignore the time factor during the process of mining the theme,which causes the result is not accurate.In order to solve this problem,TIF-LDA micro-blog theme model used variable time window is proposed to make time restrict for Micro-blog theme analysis,add time weight to micro-blog entry based on the micro-blog publishing time,use the sum of time weights as the entry′s impact factor in the LDA theme mining.The experimental results show that compared with the standard LDA theme model,the proposed model can reflect the user＇s latest focus more accurately.
作者
冯勇
屈渤浩
徐红艳
王嵘冰
FENG Yong;QU Bohao;XU Hong-yan;WANG Rong-bing(School of Information,Liaoning University,Shenyang 110036,China)
出处
《小型微型计算机系统》
CSCD
北大核心
2018年第9期2067-2071,共5页
Journal of Chinese Computer Systems
基金
辽宁省档案科技项目(L-2016-8-7)资助
辽宁省博士科研启动基金项目(201601099)资助
2016年省级本科教改立项一般项目(201607)资助
