A Knowledge-Based Tutor for Music Composition. CITE Report No. 16.

The work described here forms part of a project using models of musical ideas within an artificial intelligence and education framework whose goal is to encourage and facilitate music composition by novices. Formal knowledge of the domain (popular music) is too incomplete and fragmented to support a traditional expert-based tutor for precisely constrained tasks with clear-cut rules. Instead the proposed system will try to construct teaching plans on the basis of what novices can do, and equally importantly, what they like, in order to help them find paths to personally important goals. As well as an attempt to tackle the stringent demands of the particular domain, the research is an exploration of whether it is possible to tutor effectively with incomplete knowledge in a complex domain. A 20-item reference list is included. (Author/DB)