CTS2-LE Interoperability

CTS2

About CTS2

The Common Terminology Services Version 2 is a joint standardization effort from HL7 and OMG. The objective was to provide a model and specification for discovering, accessing, distributing and updating terminological resources. While HL7 defined the functional model for managing terminology artifacts, OMG mapped this abstract model onto a concrete information model and service interfaces, which again can be bound to various standards (e.g. SOAP, REST, RDF). Even though CTS2 was developed with healthcare use cases in mind, it can as well be used for managing and distributing terminology resources from other domains.

For further information on CTS2 see:

• OMG CTS2 Homepage (contains links to all published versions of the CTS2 standard)
• Introduction to CTS2 (Presentation from Harold Solbrig, Mayo Clinic)

Use of CTS2 in CTS2-LE

CTS2-LE's internal data model is an RDF-Binding to version 1.0 of the CTS2 standard. By this CTS2-LE supports all terminology resources defined in the CTS2 standard and follows the CTS2 service functional model for discovering and sharing these resources. The full RDF mapping onto the CTS2 model can be found here.

Interoperability with 3rd-Party Solutions

CTS2-LE can manage any semantic resource that complies to the CTS2 functional model. This not only includes terminologies but even networked ontologies and structured knowledge bases. Owners of such semantic resources can easily load these into CTS2-LE and by this utilize the standard SPARQL and REST interfaces of CTS2-LE to provide users access to these resources.

RDF

About RDF

Modern semantic (web) technologies have its root in mature knowledge representation (KR) methods and techniques. They can be seen as a “Webification” of KR languages such as the Frame Language and Description Logics (DL). In fact, the Resource Description Language (RDF) standard can be considered as a simple frame language as well as a language for semantic nets and OWL has its direct foundation in a certain DL dialect. The semantics of these languages are the theoretical backbone of controlled vocabularies and are widely used to define concrete vocabularies.

For instance, SNOMED-CT has the expressivity of the OWL EL++ dialect and even LOINC can be proper represented and processed by means of DL. Moreover, methods and techniques beyond representation such as DL-inference, logical rules , and querying via SPARQL are the building blocks for processing vocabularies.

Another important aspect is the utilization of RDF for storing and querying linked (big) data. So far many RDF stores (Jena TDB, Virtuoso, etc.) are available and able to hold data in the range of several hundred million triples. Compared to object stores and assuming an average number of 10 object attributes, one can easily store and retrieve several million objects. These capabilities would give us the opportunity to “weave” instance nets for arbitrary medical data objects together with referenced vocabularies.