Overcoming an obstacle in expanding a UMLS semantic type extent

Document Type

Article

Publication Date

2-1-2012

Abstract

This paper strives to overcome a major problem encountered by a previous expansion methodology for discovering concepts highly likely to be missing a specific semantic type assignment in the UMLS. This methodology is the basis for an algorithm that presents the discovered concepts to a human auditor for review and possible correction. We analyzed the problem of the previous expansion methodology and discovered that it was due to an obstacle constituted by one or more concepts assigned the UMLS Semantic Network semantic type Classification. A new methodology was designed that bypasses such an obstacle without a combinatorial explosion in the number of concepts presented to the human auditor for review. The new expansion methodology with obstacle avoidance was tested with the semantic type Experimental Model of Disease and found over 500 concepts missed by the previous methodology that are in need of this semantic type assignment. Furthermore, other semantic types suffering from the same major problem were discovered, indicating that the methodology is of more general applicability. The algorithmic discovery of concepts that are likely missing a semantic type assignment is possible even in the face of obstacles, without an explosion in the number of processed concepts. © 2011 Elsevier Inc.

Identifier

84856382499 (Scopus)

Publication Title

Journal of Biomedical Informatics

External Full Text Location

https://doi.org/10.1016/j.jbi.2011.08.021

ISSN

15320464

PubMed ID

21925287

First Page

61

Last Page

70

Issue

1

Volume

45

Grant

3R01LM008445-03S1

Fund Ref

U.S. Department of Health and Human Services

This document is currently not available here.

Share

COinS