We have seen the emergence of a generation of medical terminologies satisfying systematic inheritance of relationships. Examples include SNOMED CT (the Systematized Nomenclature of Medicine - Clinical Terms), the National Cancer Institute Thesaurus (NCIT), Kaiser's Convergent Medical Terminology (CMT), the VA's internal enterprise terminology, the Foundational Model of Anatomy (FMA), and the Medical Entities Dictionary (MED). These terminologies are of substantial size and complexity. Due to this, user orientation is difficult, especially given the fact that more knowledge is continually being added to them. Orientation and navigation capabilities are essential for effective terminology maintenance and usage (for example, in decision-support systems, patient records, and healthcare administrative systems). One cannot reasonably be expected to maintain or use a terminology reliably without them. We propose to design structural abstraction methodologies to derive novel terminological views called """"""""taxonomies."""""""" These will form the bases for new techniques to support user orientation to and navigation of terminologies. The taxonomies will further aid in efficient auditing. A number of different levels of taxonomy will be developed. Our methodologies will utilize the IS-A relationship hierarchies and accompanying systematic relationship inheritance of this generation of terminologies. They will be based on new partitioning techniques, which will break down large collections of concepts into smaller units of structurally and semantically similar concepts that can be more easily handled and comprehended. One partitioning technique will be based on similar relationship structure, leading to the derivation of an abstraction network call the """"""""area taxonomy."""""""" Another will further utilize semantic similarity based on common ancestry in the terminology's ISA hierarchy leading to the finer-grained """"""""p-area taxonomy."""""""" Additional partitioning techniques---leading to further refined taxonomies---will focus on other structural features of terminologies, such as obtainment-pattern regions. Our abstraction methodologies will be general and applicable to a wide range of terminologies satisfying systematic inheritance. We will use SNOMED and the NCIT's genomics hierarchies as test-beds. We will demonstrate the utility of our taxonomy-based methodologies by defining various complexity measures with respect to the underlying terminology networks and by tracking terminology evolution.
