next up previous
Next: Conclusion and Future Up: Towards Interoperable Heterogeneous Information Previous: Current Implementation Status

Related Work

 

Several approaches have been proposed for information gathering from multiple heterogeneous information sources, including multidatabases and federated databases, intelligent integration of information architectures [30], and standard interfaces for software integration and communication such as IBM's DSOM and Microsoft's OLE.

A conventional approach [1] for multidatabase or federated database management relies on building a single global schema to encompass the differences among the multiple local database schemas. Although the enforcement of a single global schema through data integration yields full transparency for uniform access, component databases have much restricted autonomy and scalability. Component schemas cannot evolve without the consent from the integrated schema. Recent activities in the OMG [26] and the ODMG standard [10], which extends the OMG object model to the database interoperability, are the important milestones for applying object-orientation to distributed object management.

The intelligent integration of information (I) reference framework [30,31] can be seen as a generic system architecture for information integration. Several projects are currently undergoing to adapt and extend this architecture with different emphases. Examples include the Garlic project at IBM Almaden Research Center [8], which targets at developing a system and tools for the management of large quantities of heterogeneous multimedia information, the TSIMMIS [12] at Stanford, which implements the mediators-based information integration architecture through a simple object exchange model.

Other representative examples include the Context Exchange project at MIT [14,27], which uses context knowledge in a context mediator to explicitly define the meaning of information provided by a source, and the SIMS project [3,2] which adapts the LOOM knowledge representation system as the data model and uses the LOOM query language to implement the agent-based integration.

In addition, a number of proposals have competed as the basic enabling technologies for implementing interoperable objects in distributed and dynamic object computing environments. Examples include Microsoft's Object Linking and Embedding (OLE), IBM's System Object Model (SOM) and its distributed version (DSOM), OMG's Common Object Request Broker Architecture (CORBA) [26]. Many of these are available as deployed software packages. Although these proposals are clearly practical and important, they focus primarily on the software interface problem, not the USECA properties in integration and access of multiple heterogeneous information sources. DIOM can be seen as a glue that spans and integrates these interface models at a higher level.



next up previous
Next: Conclusion and Future Up: Towards Interoperable Heterogeneous Information Previous: Current Implementation Status



Ling Liu
Thu Aug 15 17:49:43 MDT 1996