Quality Assurance of Concept Roles in the National Cancer Institute thesaurus

Abstract

Ontologies are integral components of many health information processing environments. Hence, quality assurance (QA) of the conceptual content of any ontology is essential. A methodology for facilitating the identification and correction of a kind of error called “missing-role error” for ontology concepts is presented. Specifically, being a member of the “null-role set” is used as an indicator that missing-role errors are likely. Our QA methodology was applied to the Biological Process hierarchy, one of the hierarchies of the National Cancer Institute thesaurus (NCIt), a prominent ontology in the biomedical domain. Two hypotheses pertaining to the effectiveness of our QA approach were investigated. Many missing-role errors were discovered and confirmed in our analysis. In comparison to a control sample, our QA approach yielded a statistically significant number of concepts with missing-role errors, confirming our two hypotheses. QA is a critical part of an ontology's life-cycle, and automated or semi-automated tools for supporting this process are invaluable. The presented approach is a useful addition to the arsenal of tools available to QA personnel.