AN ADAPTIVE OBJECT-ORIENTED APPROACH TO INTEGRATION AND ACCESS OF HETEROGENEOUS INFORMATION-SOURCES

A large-scale interoperable database system operating in a dynamic env ironment should provide Uniform access to heterogeneous information so urces, Scalability to the growing number of information sources, Evolu tion and Composability of software and information sources, and Autono my of participants, both information consumers and information produce rs. We refer to these set of properties as the USECA properties [29]. To address the research issues presented by such systems in a systemat ic manner, we introduce the Distributed Interoperable Object Model (DI OM). DIOM promotes an adaptive approach to interoperation via intellig ent mediation [46, 47], aimed at enhancing the robustness and scalabil ity of the services provided for integrating and accessing heterogeneo us information sources. DIOM's main features include (1) the recursive construction and organization of information access through a network of application-specific mediators, (2) the explicit use of interface composition meta operations (such as specialization, generalization, a ggregation, import and hide) to support the incremental design and con struction of consumer's domain query model, (3) the deferment of seman tic heterogeneity resolution to the query result assembly time instead of before or at the time of query formulation, and (4) the systematic development of the query mediation framework and the procedure of eac h query processing step from query routing, query decomposition, paral lel access planning, query translation to query result assembly. To ma ke DIOM concrete, we outline the DIOM-based information mediation arch itecture, which includes important auxiliary services such as domain-s pecific metadata library and catalog functions, object linking databas es, and associated query services. Several practical examples and appl ication scenarios illustrate the flavor of DIOM query mediation framew ork and the usefulness of DIOM in multi database query processing.