Improving adaptation in web-based educational hypermedia by means of knowledge discovery

Most adaptive web-based hypermedia systems adapt presentation of the content and/or navigation using predefined set of rules. Considering different behavior and preferences of each user it may be hard to generalize and construct all appropriate rules in advance. This problem is more noticeable in educational adaptive hypermedia systems, where adaptation to individual learning style of a student is important for the student to effectively assess particular domain. In this paper we present techniques for data mining, which can be used to discover knowledge about students' behavior during learning, as well as techniques, which take advantage of such knowledge to recommend students lessons they should study next. We also describe a process of recommendation based on knowledge discovery and present an architecture of a web-based system, which uses proposed approach to improve adaptation. Proposed architecture is independent of actual adaptive hypermedia system used.