Usage of machine learning to help in browsing the Web

Dunja Mladenic

The paper describes design of {\bf Personal WebWatcher}, the system that uses some of the known machine larning techniques to help user browsing World Wide Web. The system observes users of World Wide Web and suggests pages they might be interested in. It learns user interests from the pages requested by the user. The learned model of user interests is then used to suggest hyperlinks on new HTML-documents requested by and presented to the user via Web browser that enables connection to ``proxy'' eg. Netscape.