整合中文维基语义的网络论坛话题追踪方法研究

Research on Topic Tracking Method for Web Forums by Integrating Chinese Wikipedia Semantics

研究中文维基语义图上的网络论坛话题追踪，对于提高舆情分析系统获取情报的效率及信息处理质量有很大价值。以涉军主题帖子为背景，借鉴维基百科语义相关度与词语共现关系，构建反映词语间静态与动态关联的文本概念图，改进PR算法的关键词挖掘方法，利用维基知识解决论坛文本中的语义特征稀疏问题，减少噪音以提高论坛文本语义相关度计算的准确性。最后实验证明该方法的优越性。

It is very important for improving efficiency and quality of the public opinion analysis system to study topic tracking on Chinese Wikipedia semantic. Based on topics related to military, Wikipedia semantic correlation with word co- occurrence relations has been introduced. Text concept Graph to reflect the words between the static and dynamic has been constructed. Two ways to improve the PR algorithm of mining the keywords are suplied. The problem of the thin semantic features method using the wiki knowledge to is solved. It is well to reduce the noise in order to improve the accuracy. It is proved of the superiority of the method for the experiment.