Abstract
Location-based recommender systems in the tourist domain are increasingly becoming more and morepopular. However, these systems typically suffer from two problems: Not having sufficient information aboutthe user results in the cold start problem; and acquiring suitable user models can be problematic, the knowledgebottleneck problem. The work presented here demonstrates how stereotype modelling can be used as a suitabletool for acquiring knowledge and building user models for a Bayesian network based recommender system.