An efficient query evaluation in a mediator based on implementation plan

Recent research which purports to integrate distributed information has been concentrated on developing efficient mediator systems that not only provide a high degree of autonomy for local users, but also support the flexible integration of required functions for global users. However, there has been little attention paid to the implementation of the global views defined in a mediator. It is possible to speed up the execution of a global query if the previous result of a view is materialized in a mediator. Since the integration schema of a mediator can be incrementally modified and its usage can also be continuously varied, view usage should be carefully monitored to determine the optimized set of materialized views. Furthermore, as the number of views increases, the optimization process itself may take too long, so that the optimized set identified by a long optimization process becomes obsolete due to the recent change of view usage. This paper proposes the adaptive selection of a practical implementation method for each global view, such that available storage in a mediator can be highly utilized at any time. In order to differentiate the recent usage of a view from the past, the accumulated usage frequency of a view decays as time goes by. As a result, it is possible to change the implementation of each view adaptively with respect to the recent variation of view usage.