基于主题词的微博热点话题发现  

HOT MICROBLOGGING TOPICS DISCOVERY BASED ON SUBJECT TERMS

作　　者：叶成绪[1,2] 杨萍[3] 刘少鹏[2] 

机构地区：[1]青海师范大学计算机学院,青海西宁810008 [2]中山大学信息科学与技术学院,广东广州510006 [3]青海师范大学生命与地理科学学院,青海西宁810008

出　　处：《计算机应用与软件》2016年第2期46-50,共5页Computer Applications and Software

基　　金：国家社会科学基金项目(13BXW037);教育部春晖计划项目(Z2011023)

摘　　要：近年来,微博网站已成为海量信息的发布平台。微博丰富的信息为用户提供便利的同时,也带来了信息过载的风险。针对热点话题发现能够降低信息过载的风险,改善用户体验。结合最长公共子串和维基百科知识,提出一种基于主题词的中文微博热点话题发现方法。首先,获取微博数据的高频最长公共子串,作为描述话题的候选主题词;其次,利用维基百科知识,对候选主题词进行筛选;最后,对主题词集合聚类以发现话题,并计算每个话题的能量,从中选取热点话题。在真实数据集上的实验表明,该方法能有效发现微博热点话题。In recent years, microblogging websites have become the publishing platform of massive information. While providing conven- ience to users, the abundant microblogging information also brings in the risk of information overload. Hot topics discovery can reduce the risk of information overload and improve user experience. Aiming at this, in this paper we present a subject terms-based hot topics discovery meth- od for Chinese microblogging in combination with longest common substrings and Wikipedia knowledge. First, it acquires the high-frequency longest common substring of microblogging as candidate subject terms of description topics. Secondly, it utilises Wikipedia knowledge to screen candidate subject terms. Finally, it collects and clusters the subject terms to discover the topics, and calculates the energy of each top- ic and then selects the hot topics among them. Experiment conducted on real dataset demonstrate that our method can effectively discover hot microblogging topics.

关 键 词：主题词 维基百科 最长公共子串 热点话题发现 微博 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]