Affiliations: Universidad Católica San Antonio, Guadalupe, Murcia 30107, Spain. E-mail: {acaballero,amunoz,jsoto}@ucam.edu | University of Middlesex, UK and Universidad de Murcia, Spain
Note: [] This work is partially supported by the Spanish national research project TIN2011-28335-C02-02, Foundations for the Development of AAL Services and Applications, through the Bureau of Innovation and Economy, and the research project PMAFI/26/12, founded by the Catholic University of Murcia (UCAM).
Note: [] Corresponding author.
Abstract: This paper offers an agent-based mechanism in intelligent environments (IE) to recommend profiles to users according to their requirements. This mechanism is based on similarity among profiles as well as trust and reputation among agents. Our proposal is employed in an intelligent university campus where there are different activities that require user profiling (e.g., management of resources such as classrooms, meeting rooms, computers in labs, etc). In these activities, a user profile can be seen as a set of recommendations on resources configurations or services that fulfill users' requirements. The proposed mechanism uses an ontology to model the basic concepts related to the profile assignment process. The ontology is provided along with two extensions for representing two specific application domains at the university campus: assignment of meeting rooms and configuration of virtualized environments for iLabs. The performance of the proposed mechanism is evaluated using a university campus scenario simulated by means of a localization network and a prototype application. The results presented in this paper are collected during the interactions between two types of agents (user and broker) aimed to select the most adequate profiles in each situation. Broker agents estimate the suitability of a profile using trust and reputation information coming from its own experiences or from other brokers. Under uncertainty conditions, arguments offered by broker agents and similarity between two set of user's requirements and preferences are also taken into account to select the best profile.
Keywords: User profiling, intelligent environment (IE), ontology, trust and reputation, argumentation
