Abstract: In the future, universal multimedia access will become the mainstream service pattern of multimedia service via networks. Aiming at the access pattern, the MPEG21 proposed the digital item adaptation (DIA) technique archi- tecture to standardize the related techniques. However, DIA specifics some metadata such as context description only. The key adaptation decision-taking engine is still an open question left. Considering the shortcomings of existing adap- tation decision models,the paper proposed a model named adaptation decision-taking tree (ADT) which looks upon ad- aptation decision as sequence decision and the decision tree is employed to model the decision process. To meet the per- sonalized demand in multimedia services, the AI)丁was constructed by user preferences. In addition, by using optimal model to create the decision space, which deduces the numbers of decision candidate, were deduced, and the decision time was cut observably, and the demand of real time was satisfied. Based on simulation, it can be concluded that the result of AI)丁is better than or similar to that of existing models in decision error and average iteration.

Key words: Adaptation decision-taking engine, Adaptation decision-taking tree, Digital item adaptation, Multimedia services