Recommendation of Multimedia Content in Digital Convergence Environments: TV and Web

The emerging scenario of interactive Digital TV (iDTV) is promoting the increase of interactivity in the communication process and also in audiovisual production, thus raising the number of channels and resources available to the user. This reality makes the task of finding the desired content becoming a costly and possibly ineffective action. The incorporation of recommender systems in the iDTV environment is emerging as a possible solution to this problem. This work aims to propose a hybrid approach to content recommendation in iDTV, based on data mining techniques, integrated to the Semantic web concepts, allowing structuring and standardization of data and consequently making possible sharing of information, providing semantics and automated reasoning. For the proposed service it is considered the Brazilian Digital TV System (SBTVD) and the middleware Ginga. A prototype has been developed and experiments carried out with NetFlix database. As results, it was obtained an average accuracy of 30% using only the data mining technique. On the other hand, the evaluation including semantic rules obtained an average accuracy of 35%.