Enhancing the Quality of Hierarchic Relations in the National Cancer Institute Thesaurus to Enable Faceted Query of Cancer Registry Data

Abstract

PURPOSE To audit and improve the completeness of the hierarchic (or is-a) relations of the National Cancer Institute (NCI) Thesaurus to support its role as a faceted system for querying cancer registry data. METHODS We performed quality auditing of the 19.01d version of the NCI Thesaurus. Our hybrid auditing method consisted of three main steps: computing nonlattice subgraphs, constructing lexical features for concepts in each subgraph, and performing subsumption reasoning with each subgraph to automatically suggest potentially missing is-a relations. RESULTS A total of 9,512 nonlattice subgraphs were obtained. Our method identified 925 potentially missing is-a relations in 441 nonlattice subgraphs; 72 of 176 reviewed samples were confirmed as valid missing is-a relations and have been incorporated in the newer versions of the NCI Thesaurus. CONCLUSION Autosuggested changes resulting from our auditing method can improve the structural organization of the NCI Thesaurus in supporting its new role for faceted query.

Detecting missing IS-A relations in the NCI Thesaurus using an enhanced hybrid approach

“…In our previous work [ 28 ], we developed a lexical-based approach to identify missing IS-A relations in the NCI Thesaurus. The lexical feature used in that work was the enriched bag-of-words (i.e., in this paper).…”

“…EVS experts confirmed 73 out of 253 suggested missing IS-A relations. We compared our hybrid approach in this paper with the lexical-based approach in previous work [ 28 ] in two aspects.…”

“…In our previous work [ 28 ], only one type of lexical features (enriched bag-of-words) was used. Therefore, in the second experiment, we tried to consider associative roles and roots of noun chunks as additional subsumption testing (i.e., in addition to lexical features and other restrictions used in [ 28 ]) to further check their effectiveness in helping identify missing IS-A relations.…”

“…In our previous work [ 28 ], only one type of lexical features (enriched bag-of-words) was used. Therefore, in the second experiment, we tried to consider associative roles and roots of noun chunks as additional subsumption testing (i.e., in addition to lexical features and other restrictions used in [ 28 ]) to further check their effectiveness in helping identify missing IS-A relations. Given 253 missing IS-A relations identified in our previous work, 135 out of 253 were the cases in which both the subconcept and the superconcept contained at least one associative role and their associative roles were not identical.…”

Informatics Tools for Cancer Research and Care: Bridging the Gap Between Innovation and Implementation

“…It includes tools that support cancer research at molecular, 3-6a cellular, 7 tissue, 8-10 organ, [11][12][13][14][15] individual, [16][17][18][19] and population [20][21][22][23] levels. The tools described also support a range of cancer informatics and data science functions, including data integration, 13,[24][25][26][27][28][29] data curation, 30 deep learning, 9 information retrieval, 20,31,32 natural language processing, 22 and statistical analysis. 5,6 A catalog of all available ITCR tools can be found on the ITCR website 33…”