This paper describes a multiagent recommender system where agents maintain local knowledge bases and, when requested to support a travel planning task, they collaborate exchanging information stored in their local bases. A request for a travel recommendation is decomposed by the system into sub tasks, corresponding to travel services. Agents select tasks autonomously, and accomplish them with the help of the knowledge derived from previous solutions. In the proposed architecture, agents become experts in some task types, and this makes the recommendation generation more efficient. In this paper, we validate the model via simulations where agents collaborate to recommend a travel package to the user. The experiments show that specialization is useful hence providing a validation of the proposed model.
展开▼
