A Multi-agent Smart User Model for Cross-domain Recommender Systems
Gustavo Ramírez-González, Beatriz López · 2005
This paper describes our approach to the next generation of open, distributed and heterogeneous recommender systems using Smart User Models (SUM). Our work focuses on integrating multiple agent-based services based on a unique representation of the user in what is called a Multi-agent Smart User Model. Intelligent agents are used in order to obtain a single user model instead of having several versions of the same user spread throughout various services. A methodology has been developed using incremental aggregation of information, which favors nonintrusive behavior of the user model in order to determine objective, subjective and emotional user features.