Semantic Integration is the process through which the items in a UML model are made consistent with similar items from other organizations' UML models, the idea being that if components of one system use the same terms as similar components of other systems, those systems will be able to easily share information, eliminating significant intervention or additional coding.

Semantic Integration is done using several tools, including:

• Semantic Integration Workbench 
• Semantic Connector 
• UML Loader

The Semantic Integration Workbench (SIW) is a Java Web Start application within the caCORE toolset that assists users in adding consistent metadata to their UML model, or verifying consistency, by matching their concepts with similar items from the NCI Thesaurus. This process helps bring the UML model semantically in-line with other registered caCORE-like systems, allowing for easier data-sharing between distributed systems.

The Semantic Connector feature of the SIW performs a search of EVS to find elements similar to those present in the UML model.  When a similar item is found, the system attaches the appropriate EVS concepts to the elements in the UML model. This provides a kind of "pre-mapping" of the UML model items, allowing for easier review and confirmation of the mapped elements.

In addition to providing a method for looking up similar information from the NCI Thesaurus, the SIW also allows users to look up and (if matches are found) bind their UML domain model objects to content that has already been registered in caDSR.

The end result of the Semantic Integration process is an annotated XMI file that can then be used to register metadata into caDSR using the UML Loader. The caDSR acts as the central repository for all registered metadata and makes that metadata available for other systems to use or map to.

The UML Loader is a tool used by the NCICB staff to register or upload a submitted system's metadata (provided through the annotated XMI file) into the caDSR.  This process reuses existing items from the caDSR or creates new items as appropriate.

Ultimately the goal of Semantic Integration and Metadata Registration are to ensure that disparate and distributed clinical data repositories and research systems are using common terms for their corresponding data elements, allowing them to more efficiently and effectively share information.