Summary: | 碩士 === 銘傳大學 === 資訊管理學系碩士在職專班 === 91 === At present, Implicit Rating is generally used as a page rating schema for web browsing recommendation. They are usually combined with data mining techniques to put the browsing learning behaviors into page association rules, based on which page recommendation can be provided. Whereas, it often happened that there is not much match between the browsing activity of a student and browsing rules, and hence, the system can’t make any recommendation. In this paper, we try to exploit the bookmarks hidden in relationship among web pages. Therefore, we propose a method to calculate the similarity among bookmarks, and then use the similarity measurement to make Clustering Analysis of bookmarks. The effect of Clustering along with the outcome of Association-Rule Mining will introduce more new rules that can be used as reference for recommendation.
In this thesis, we will provide a formula for the similarity among web pages, and utilize CAST (Cluster Affinity Search Technique) approach to make Clustering Analysis of web pages. It is obvious that the outcome of such Clustering accompanied with the effect of recommendation based on the analysis of Association Rules greatly promote the Recall Rate and Precision Rate.
|