Authors:
Terrence Mason
and
Ramon Lawrence
Affiliation:
Iowa Database and Emerging Applications Laboratory, Computer Science, University of Iowa, United States
Keyword(s):
integration, database, schema, metadata, annotation, evolution, dynamic, JDBC, conceptual, embedded.
Related
Ontology
Subjects/Areas/Topics:
Coupling and Integrating Heterogeneous Data Sources
;
Databases and Information Systems Integration
;
Enterprise Information Systems
Abstract:
Current integration techniques are unsuitable for large-scale integrations involving numerous heterogeneous data sources. Existing methods either require the user to know the semantics of all data sources or they impose a static global view that is not tolerant of schema evolution. These assumptions are not valid in many environments. We present a different approach to integration based on annotation. The contribution is the elimination of the bottleneck of global view construction by moving the complicated task of identifying semantics to local annotators instead of global integrators. This allows the integration to be more automated, scaleable, and rapidly deployable. The algorithms are packaged in an embedded database engine contained in a JDBC driver capable of dynamically integrating data sources. Experimental results demonstrate that the Unity JDBC driver efficiently integrates data located in separate data sources with minimal overhead.