A key benefit of case-based reasoning (CBR) and recommender systems is the use of past experience to guide the synthesis or selection of the best solution for a specific context or user. Typically, the solution presented to the user is based on a value system that privileges the closest match in a query and the solution that performs best when evaluated according to predefined requirements. In domains in which creativity is desirable or the user is engaged in a learning activity, there is a benefit to moving beyond the expected or "best match" and include results based on computational models of novelty and surprise. In this paper, models of novelty and surprise are integrated with both CBR and Recommender Systems to encourage user curiosity.
展开▼
