Personalization of a hybrid integration system: Creation of views to materialize based on the distribution of user queries

The hybrid mediator is an integration system where one part of data is queried on demand as in the virtual approach, while another part is extracted, filtered and stored in a local database. The selection of data to be stored in the mediator (to materialize) is an essential task. Different selection criteria were proposed to select among a set of candidate views, ones that will be materialized. However, the approaches proposed with regard to the creation of these views are rare. This paper presents a data materialization approach, in the scope of a hybrid integration system. This approach creates candidate views for materialization based on the previous user queries, and then decides which among them should be selected for materialization. Guided by the technique of clustering-merging, it is organized into two steps: (1) from the distribution of previous user queries, are extracted all data most queried by users. The obtained data are classified as candidate views for materialization. (2) then selecting which views to materialize among all the candidate views created in the first step.