首页 | 本学科首页   官方微博 | 高级检索  
     

基于反馈报道的话题模型动态修正方法
引用本文:郑燕,鲁燃,赵爱华. 基于反馈报道的话题模型动态修正方法[J]. 计算机应用, 2012, 32(5): 1343-1346
作者姓名:郑燕  鲁燃  赵爱华
作者单位:1. 山东省分布式计算机软件新技术重点实验室,济南 2500142. 山东师范大学 信息科学与工程学院,济南 250014
基金项目:国家自然科学基金资助项目(60873247);山东省高新自主创新专项工程(2008ZZ28);山东省自然科学基金资助项目(ZR2009GZ007);济南市青年科技明星计划项目(20080201)
摘    要:在话题追踪过程中,由于给定的初始话题相关报道少,而且话题具有动态演变的特点造成话题模型不准确。针对这一问题,提出了利用动态阈值收集反馈报道构造话题修正模型,实现了话题模型的动态修正;同时结合命名实体能够更加有效地区分不同话题的特性,提出了在修正话题模型时增大相关命名实体权重的方法,从而获得更准确的话题表示模型。实验结果表明,该方法能有效避免话题漂移现象,降低话题追踪过程中的漏报率和错报率。

关 键 词:话题追踪  话题模型  动态阈值  命名实体  反馈报道  
收稿时间:2011-10-28
修稿时间:2011-12-16

Dynamic topic model amending method based on feedback stories
ZHENG Yan , LU Ran , ZHAO Ai-hua. Dynamic topic model amending method based on feedback stories[J]. Journal of Computer Applications, 2012, 32(5): 1343-1346
Authors:ZHENG Yan    LU Ran    ZHAO Ai-hua
Affiliation:1. School of Information Science and Engineering, Shandong Normal University, Jinan Shandong 250014, China
2. Shandong Provincial Key Laboratory for Distributed Computer Software Novel Technology, Jinan Shandong 250014,China
Abstract:In topic tracking,the initial topic related stories are few and topic evolves dynamically,which leads to the topic model could not express topic accurately.Concerning this problem,it was proposed to build amended topic model by feedback stories collected by dynamic threshold,to amend topic model dynamically.And in combination of the feature that the named entity could differentiate different topics more effectively,it was suggested to increase the weight of named entity when amending topic model,to express a topic better.The experimental results indicate that,this method can solve the topic shifting problem effectively,and the miss tracking rate and fault tracking rate decrease a lot in topic tracking.
Keywords:topic tracking  topic model  dynamic threshold  named entity  feedback stories
本文献已被 CNKI 万方数据 等数据库收录!
点击此处可从《计算机应用》浏览原始摘要信息
点击此处可从《计算机应用》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号