caCORE SDK Workflow. A UML object model and a description-logic terminology source (NCI Thesaurus in the present work) are the inputs into the workflow. The model is exported from the format native to the tool it was developed in to the standard XMI representation. The XMI file is then annotated with terminology concepts using Semantic Connector. The annotated model is used as input into Codegen, which generates the software for a caBIG-compatible data system with object-oriented APIs. The annotated model is loaded as metadata into the caDSR using UML Loader. Model metadata is reviewed and completed by a curator using caDSR utilities, and then becomes available from the caDSR APIs and web applications.