面向网络论坛的突发话题发现

发布时间：2018-03-30 18:09

本文选题：计算机应用　切入点：中文信息处理　出处：《中文信息学报》2010年03期

【摘要】：每天有大量的信息涌现在论坛上,用户可以通过论坛获知目前国际国内正在发生的一些突发事件。如何使用机器自动化的方法检测论坛中的突发话题已经成为搜索引擎以及网络挖掘系统的一项基础任务。话题检测与跟踪模型(TDT)可以很好的解决话题发现问题,但是TDT处理的对象是新闻语料,与论坛内容相比,新闻语料更准确、严谨、规范。TDT中使用的方法不适合用语随意的论坛。因此在网络论坛这种噪音环境下的话题检测面临着一定的困难与挑战。文中提出一种基于噪音过滤的话题发现模型,它从内容和用户参与度两个角度来检测论坛话题。在"水木社区"的"水木特快"上进行了相关的实验,实验结果表明该文提出的模型不仅可以检测突发话题,而且可以检测与这些话题相对应的用户社区。
[Abstract]:Every day there is a lot of information coming up in the forum, Users can be informed through the forum of the current international and domestic events that are happening. How to use the method of machine automation to detect unexpected topics in the forum has become a basis for search engines and web mining systems. Topic Detection and tracking Model (TDTT) is a good way to solve the problem of topic discovery. But TDT deals with the news corpus, which is more accurate and rigorous than the content of the forum. The method used in the specification .TDT is not suitable for a random forum. Therefore, topic detection in a noisy environment such as a network forum faces some difficulties and challenges. In this paper, a topic discovery model based on noise filtering is proposed. It detects forum topics from the perspective of content and user participation. Experiments are carried out on Waterwood Express in "Waterwood Community". The experimental results show that the model presented in this paper can not only detect unexpected topics. And you can detect user communities that correspond to these topics.
【作者单位】：中国科学院计算技术研究所;中国科学院研究生院;
【基金】：国家自然科学基金重点项目资助(60933005) 国家973基础研究计划资助项目(2007CB311100) 国家863计划资助项目(2007AA01Z438)
【分类号】：TP393.094

【共引文献】