Mining on Dynamic Web Pages by Inductive Logic Programming

• Main Authors: Ko-Chang Chang, 張克彰
• Other Authors: 吳志宏
• Format: Others
• Language: zh-TW
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/86458099249823660716

碩士 === 樹德科技大學 === 資訊管理研究所 === 91 === Web usage mining is a means to analyze the browsing behaviors of users on the Internet. This research employs the techniques of inductive logic programming (ILP) to explore web usage mining with dynamic web pages. Since dynamic web pages are generated by executing CGI-programs with a series of parameters, the relationships among the parameters can be viewed as the background knowledge to ILP. We collect from the web logs and user profiles and classify them into positive and negative examples as inputs to ILP accordingly. Hypotheses of the browsing patterns are automatically generated by ILP. The benefits of using ILP on web usage mining include (1) readable and structured representation of the background knowledge about the web pages and users; (2) reasonable and explainable conclusions supported by inductive learning theories. In this thesis, we target on identifying the customization problem of web pages and the reduction of communication bandwidth caused by dynamic web pages. Experimental results show that ILP can serve as flexible platform for web usage mining.