A Rule-Learning Approach for the Personalization of Context-Aware Multimedia Documents Adaptation Processes

The execution of multimedia documents in pervasive environments may involve the adaptation of their contents according to users context information. In this regard, many adaptation approaches have been proposed. Generally, these approaches deals with context collection, representation and interpretation without taking into account its storage and analysis. This feature helps provide prior knowledge about the past decisions that users may take in the future. Thus, we propose in this paper a rule-learning approach for the personalization of adaptation rules within context-aware multimedia documents adaptation processes. The proposal uses rule based data mining classifiers, by means of the sequential covering algorithm. Our proposal is validated through scenarios implemented in a real prototype that systematically records the context values and the actions performed accordingly. These data, are then used by the rule learning algorithm in order to personalize the adaptation actions to users. The obtained results are very satisfactory and encouraging.