A Validated Ontology for Metareasoning in Intelligent Systems

Manuel F Caro; Michael T Cox; Raúl E Toscano-Miranda

doi:10.3390/jintelligence10040113

A Validated Ontology for Metareasoning in Intelligent Systems

J Intell. 2022 Nov 24;10(4):113. doi: 10.3390/jintelligence10040113.

Authors

Manuel F Caro¹, Michael T Cox², Raúl E Toscano-Miranda¹

Affiliations

¹ Education, Technology & Language (EduTLan Research Group), Department of Educational Informatics, University of Córdoba, Carrera 6 No. 77-305, Montería 230002, Córdoba, Colombia.
² Education, Department of Computer Science & Engineering, College of Engineering and Computer Science, Wright State University, Dayton, OH 45324, USA.

Abstract

Metareasoning suffers from the heterogeneity problem, in which different researchers build diverse metareasoning models for intelligent systems with comparable functionality but differing contexts, ambiguous terminology, and occasionally contradicting features and descriptions. This article presents an ontology-driven knowledge representation for metareasoning in intelligent systems. The proposed ontology, called IM-Onto, provides a visual means of sharing a common understanding of the structure and relationships between terms and concepts. A rigorous research method was followed to ensure that the two main requirements of the ontology (integrity based on relevant knowledge and acceptance by researchers and practitioners) were met. The high accuracy rate indicates that most of the knowledge elements in the ontology are useful information for the integration of multiple types of metareasoning problems in intelligent systems.

Keywords: heterogeneity problem; intelligent systems; metareasoning ontology; metareasoning problem; ontology validation.

Grants and funding

This research was funded in part by NSF grant number 1849131 and by the Office of Naval Research grant number N00014-18-1-2009. This research was funded in part by University of Córdoba—Colombia.