Syskill和Webert网页评级数据集

Michael Pazzani Department of Information and Computer Science, University of California, Irvine Irvine, CA 92697-3425 pazzani '@' ics.uci.edu http://www.ics.uci.edu/~pazzani Data Set Information: The HTML source of a web page is given. Users looked at each web page and inidated on a 3 point scale (hot medium cold) 50-100 pages per domain. However, this is realistic because we want to learn user profiles from as few examples as possible so that users have an incentitive to rate pages. Attribute Information: Each subject is in a separate directory. Within each directory, there is an file named "index". The index contains information on the other files. Each entry is a line of the form: file-name | rating | url | date-rated | title where file-name is the name of a file (usually an integer), rating is hot, medium, or cold. There are so few medium's that mediums are usually merged with cold in experiments. The other fields aren't used in learning, but they are collected by the interface for other purposes. They are the url of the html source, the date rated and the title of the web oage. Relevant Papers: Pazzani M., Billsus, D. (1997). Learning and Revising User Profiles: The identification of interesting web sites. Machine Learning 27, 313-331 [Web link] Pazzani, M., Muramatsu J., Billsus, D. (1996). Syskill & Webert: Identifying interesting web sites. Proceedings of the National Conference on Artificial Intelligence, Portland, OR. PDF [Web link]