This paper reports on our approach towards a federated database system. The tools and techniques developed in our project for semantic integration assist incremental integration driven by actual information requests of end users and the dynamic maintenance of integrated schemas driven by external schema evolution. This approach tries to meet the requirements of realistic situations with a large number of external information bases (which - because of their autonomy - are subject only to locally controlled constant change), in which completely integrated views valid for all users can hardly be achieved with reasonable effort. In addition, many available information sources (e.g., online databases) do not even provide fine-grained, explicitly structured data as relational databases do. Thus, in our approach we also develop tools for the structural enrichment of data from external information sources that do not provide any kind of schema. This paper in particular focuses on a declarative methodology for schema integration as opposed to the procedural} approach of supporting a number of schema transformation operations to overcome inconsistencies or supporting just a multidatabase query language as advocated by the loosely coupled multidatabase approach. We aim at integrated schemas to which heterogeneous schemas are mapped to resolve differences between corresponding subschemas. However, we do not aim at a complete, global integrated schema, which overcomes all heterogeneities, but rather want to enhance the incremental design and maintenance of integrated schemas, according to the specific needs of an integrated application.

Top