基于带权语义距离的网页预取方法  

Web Prefetching Method Based on Weighted Semantic Distance

作　　者：赵欣欣[1] 索红光[1] 刘玉树[1] 张利萍[1] 

机构地区：[1]北京理工大学计算机科学技术学院,北京100081

出　　处：《北京理工大学学报》2006年第8期708-711,共4页Transactions of Beijing Institute of Technology

基　　金：国家部委预研项目(40104116)

摘　　要：为了体现词条之间的关系,更好地为用户提供个性化服务,提出了在语义网络下对用户兴趣建模的思想.根据用户的访问历史,利用知网建立基于概念关系的用户兴趣森林模型.通过计算机链接描述文字的平均带权语义距离,预取用户将来可能访问的Web页.基于带权语义距离的网页预取方法对用户兴趣进行归纳,用带权语义距离反映词与词之间的概念关系.实验结果表明,此方法的命中率较高,漏取率较低,平均命中率可以达到61%左右.To embody the relations between terms and provide a better personalized service for the user, the idea of constructing the user interest model under the semantic network is proposed. According to the user＇s navigation history and the concept relations living in the HowNet, a user interest forest model based on the relative relations of concepts is constructed. By computing the average weighted semantic distance of texts in the hyperlinks, Web pages that a user may visit in the future are prefetched. The Web prefetching method based on weighted semantic distance induces the user interests and reflects the concept relations of the terms by weighted semantic distance. Experiments showed that this method had a higher hit-ratio and a lower miss-ratio, and the average hit-ratio was about 61 percent.

关 键 词：网页预取 带权语义距离 用户兴趣森林 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]