ITcon Vol. 19, pg. 474-493, http://www.itcon.org/2014/28

Validating ontologies in informatics systems: approaches and lessons learned for AEC

revised:July 2012
published:October 2014
editor(s):Rezgui Y
authors:Tamer E. El-Diraby, Associate Professor,
University of Toronto;
tamer@ecf.utoronto.ca, www.i2c.ca
summary:In their pursuit to represent a human-savvy machine interpretable model of knowledge, informatics ontologies span three dimensions: philosophy, artificial intelligence, and linguistics. This poses several challenges to ontology validation. Within the scope of knowledge models, four types of validity are relevant: statistical, construct, internal and external. Based on benchmarking some tools and best practices from other domains, a map is proposed to link specify a set of tools to support the handling of four validity types (statistical/conclusion, internal, construct, and external) in each of the three dimensions. The map advocates a debate-based approach in validating the philosophical dimension to allow for innovation and discovery; use of competency questions and automated reasoning tools for the artificial intelligence dimension; and experimenting with lexical analysis tools (especially web contents) for the linguistic dimension. A set of best practices are proposed based on benchmarking other domains. These include falsifying the conceptual frameworks of research methodologies, scope management, iterative development, adequate involvement of experts, and peer review.
keywords:Informatics, Ontology, Research Validation, Knowledge Management
full text: (PDF file, 0.363 MB)
citation:El-Diraby TE (2014). Validating ontologies in informatics systems: approaches and lessons learned for AEC, ITcon Vol. 19, pg. 474-493, http://www.itcon.org/2014/28