An Effective Relevance Prediction Algorithm Based on Hierarchical Taxonomy for Focused Crawling

摘要：

How to give a formal description for a users interested topic and predict the relevance of unvisited pages to the given topic effectively is a key issue in the design of focused crawlers.However,almost all previous known focused crawlers do the Relevance Predication based on the Flat Information (RPFI) of topic only,i.e.regardless of the context between keywords or topics.In this paper,we first introduce an algorithm to map the topic described in a keyword set or a document written in natural language text to those described in hierarchical topic taxonomy.Then,we propose a novel approach to do the Relevance Predication based on the Hierarchical Context Information (RPHCI) of the taxonomy.Experiments show that the focused crawler based on RPHCI can obtain significantly higher efficiency than those based on RPFI.

关键词： Focused Crawling Relevance Prediction Hierarchical Topic Taxonomy Topic Description

作者: Zhumin Chen Jun Ma Xiaohui Han Dongmei Zhang

作者单位: School of Computer Science & Technology,Shandong University,Jinan,250061,China

会议类型: 国际会议

会议名称: 4th Asia Information Retrieval Symposium(AIRS 2008)(第四届亚洲信息检索研讨会)

会议地点: 哈尔滨

会议语种:英文

页码: 613-619

在线出版日期: 2008-01-16（万方平台首次上网日期，不代表论文的发表时间）