Negociación aplicada a Recomendación a Grupos: Un Estudio sobre Usuarios en el Dominio de Películas

CHRISTIAN VILLAVICENCIO; SILVIA SCHIAFFINO; J. ANDRES DIAZ PACE; ARIEL MONTESERIN

Córdoba

Simposio; Simposio Argentino de Inteligencia Artificial (ASAI 2017); 2017

SADIO

Providing recommendations to groups of users has become popular in many applications today. Even though there are several group recommendation techniques, the generation of recommendations that satisfy the group members in an even way remains a challenge. Because of this, we have developed a multi-agent approach called MAGReS that relies on negotiation techniques to improve group recommendations. Our approach was tested (on the movies domain) using synthetic data with satisfactory results. Given that the results when using synthetic data may sometimes differ with reality, we decided to assess MAGReS using data from real users. The results obtained showed firstly that, in comparison with the recommendations produced by a traditional approach, the recommendations of MAGReS produce a greater level of satisfaction to the group, and secondly that the proposed approach was able to predict more accurately the satisfaction levels of the group members.