Personalização para televisão digital utilizando a estratégia de sistema de recomendação para ambientes multiusuário
Lucas, Adriano dos Santos
MetadataShow full item record
The Digital Television system (TVD) increases the content offer, the audio and video quality and the possibility of services and applications when compared to traditional systems. Among the possibilities of application for this technology, we can highlight the systems able to perform recommendations of content according to the viewer s interests, theses systems are called recommendation systems. Besides offering a different personalization service, the recommendation systems can be a solution to the information overload caused by offering possible content, what makes difficult the search and localization of programs according to the viewer`s interest. Multiuser environment must be taken into account when offering the TVD viewer content personalization, that is, many viewers using the same receptor. This dissertation presents a recommendation system for multiuser environments, the RePTVD (Personalized Recommendation for Digital Television), with its architecture in the Set-top Box. The RePTVD aims at recommending content according to the behavior standards implicitly found when using the television. Therefore, information was implicitly collected and stored, data mining was performed using Apriori algorithm and finally information was filtered. A composition of theses stages was presented using a recommendation process which approaches all the necessary steps to perform the content recommendation. Due to the fact that this is a specific language to TVD and aiming the system portability, API Java TV was used to implement the proposal as a concept proof. Besides that, we could note Brazilian-standard characteristics which have Ginga as the middleware responsible for the applications performance. The evaluation was performed through a test in which data provided by IBOPE Brazilian institute was used concerning the viewing behavior from six houses collected during 14 days. The results indicated the efficiency and quality of RePTVD system implementation.