Semantic Recommender Systems for Digital TV: From Demographic Stereotyping to Personalized Recommendations

Compared to analog transmissions, Digital Television (DTV) standards allows a higher number of available TV stations and consequently, a larger entertainment offer. In this context, Recommender Systems (RS) support users in choosing entertainment content by narrowing their options to a reduced set based on their preferences an interests. However, new users or those having incomplete profiles prevent the system to produce accurate recommendations, which is more noticeable in early stages of the RS. This paper proposes the use of a demographic stereotyping approach based on minimal user attributes acquired during user registration. Furthermore, we propose an experimental procedure that can be used to compare the system accuracy for the created stereotypes and for users making extensive use of the system.