2015 IEEE International Conference on Bioinformatics and Biomedicine (BIBM) (2015)
Washington, DC, USA
Nov. 9, 2015 to Nov. 12, 2015
Christopher Ochs , Department of Computer Science, New Jersey Institute of Technology, Newark, United States
Yehoshua Perl , Department of Computer Science, New Jersey Institute of Technology, Newark, United States
James Geller , Department of Computer Science, New Jersey Institute of Technology, Newark, United States
Mark Musen , Stanford School of Medicine, Stanford University, CA, United States
Terminologies are typically large and complex knowledge systems. It is difficult to obtain an orientation into their structure and content. In previous research we designed compact summary networks called partial-area taxonomies to provide a structural summary of a terminology. The sizes of a terminology and of its partial-area taxonomy are defined as their numbers of nodes. While a partial-area taxonomy is typically smaller than the original terminology, it is often not compact enough to provide a clear "big picture," due to too many nodes that summarize only a small number of terminology concepts. The display of such a partial-area taxonomy is still overwhelming. In this paper, we introduce a more compact summary of a terminology, called an aggregate taxonomy, obtained by aggregating small partial-area taxonomy nodes into larger nodes. We present a parametrized technique to study the design of such an aggregate taxonomy and apply it to the Specimen hierarchy of SNOMED CT. A software tool for creating and displaying aggregate taxonomies is described. We illustrate how aggregate taxonomies derived across multiple SNOMED CT releases can be used to summarize the evolution of the Specimen hierarchy's content over eight years of SNOMED CT releases.
Abstraction Network, SNOMED CT, Medical Terminology and Ontology, Visualization, Summarization, Terminology Abstraction, Taxonomy
C. Ochs, Y. Perl, J. Geller and M. Musen, "Using aggregate taxonomies to summarize SNOMED CT evolution," 2015 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), Washington, DC, USA, 2015, pp. 1008-1015.