利用博客链接平台选取联合关键字的博客聚类研究

针对全文本关键字检索的时间成本高,以及采用标签/类别会产生语句歧义和同义词等问题,提出在博客链接平台上选取联合关键字进行博客聚类。假设一个博客文章被查询的候选关键字(或者联合关键字)可以用于表示这个博客文章的主题。为验证该假设,首先将跟踪代码嵌入到博客链接(BC)组件中,以收集读者查询的关键字。然后,选取适当的候选关键字作为联合关键字。最后,使用重叠投影、交互信息投影、分布式分布信息和肯德尔 系数这四种相似性度量以验证BC组件提取的联合关键字。实验结果表明,提出的方法可以为查询者提供一条找到对应博客的快速通道。此外,生成的联合关键字可以减少全文本关键字检索过程的复杂度和冗余度,很好地满足了博客用户的需求。

The Research of Blog Clustering Method by Selecting Joint Keywords on Blog Connect Platform

Concerning that the time cost of full-text keyword search time is high, and the label / category statement will produce ambiguity and synonyms problems, the way to select joint keywords in the Blog Connect platform for blog clustering is proposed. It is assumed that the candidate keywords of a blog post by the query (or joint keyword) can be used to represent the theme of this blog. In order to verify this assumption, a tracing code is embedded in Blog Connect firstly so as to collecting the keywords queried by readers. Then, FKRP is used to select candidate keywords as co-keywords. Finally, Similarity measures, including overlapping projection, mutual information projection, distributed information and the Kendall coefficient is used to validate the BC component extraction. The experimental results show that the proposed method can provide a fast channel for the query to find the corresponding blog. In addition, the joint key generation can reduce the search process complexity and redundancy, which could well meet the needs of blog users.