The COntext INterchange (COIN) Research Project

Stephane Bressan, Research Scientist
Thomas Lee, Doctoral candidate

The COntext INterchange (COIN) Research Project at MIT constitutes a suite of services being developed to support the integration of structured and semi-structured data sources by casting the relational data model over all sources which are accessible through the COIN suite.In particular, COIN focuses on the challenges presented by inconsistentdata semantics. Recognizing that native data sources or individual users often use different languages when speaking about the same topic or use similar terminology to refer to vastly different concepts, COIN introduces the notion of context to articulate assumptions about data semantics. A logic framework is employed to mediate the semantic inconsistencies between different contexts when querying or interpreting information. Because data is stored in many different kinds of systems using many different data models and query languages, a second component of the COIN service architecture is multi-database query planning and execution. Wrappers support just-in-time access to native data sources. This talk will describe the problem of semantic data integration, provide an overview of the COIN approach to the problem, and outline general COIN project objectives. After summarizing current results, the talk will turn to an open discussion, seeking opportunities for applying the COIN framework to the medical domain.