摘要
为了提高专用搜索引擎的分类精确度和可控度,提出了一种新算法。根据现有的后缀树文本聚类,通过建立基本分类表,并结合专用搜索引擎的特点进行改进,将文本分类和文本聚类有机地结合起来,使改进后的算法能够通过人工的干预,不断修正自己的分类结果。实验表明,相比传统的文本分类算法,这种算法能通过自我修正能力不断提高分类准确度,并且计算开销和传统文本分类算法相当。因此,该算法通过结合文本分类和文本聚类算法,在专用搜索引擎结果处理上提出了新的思路。
In order to promote the precision and controllability of the categorization of dedicated search engine. This paper has raised a new algorithm, which improves STC, and combines text categorization and clustering organically by building a basic clustering table. According to experiment result, this improved algorithm can continuously fix the categorization result by the manual error fixing, and cost the nearly computing time of traditional text categorization. In summary, this algorithm has raised a new thought in the result analysis of dedicated search engine by combining the text categorization and clustering.
出处
《成都信息工程学院学报》
2010年第3期269-274,共6页
Journal of Chengdu University of Information Technology
基金
教育部科学技术研究重点项目资金资助项目(107094)
关键词
计算机应用
文本分类
文本聚类
搜索引擎
后缀树
词频
降维
召回率
computer application
text categorization
text clustering
search engine
STC
term frequency
dimensionality reduction
recall rate
