Affiliations: [a] Department of Computer Science, University of Cape Town, Cape Town, South Africa. E-mail: [email protected] | [b] Council for Scientific and Industrial Research, Pretoria, South Africa. E-mail: [email protected]
Corresponding author: Zubeida Casmod Khan, Department of Computer Science, University of Cape Town, Cape Town, South Africa. E-mail: [email protected].
Note:  Accepted by: Oliver Kutz
Abstract: Modularity is being increasingly used as an approach to solve for the information overload problem in ontologies. It eases cognitive complexity for humans, and computational complexity for machines. The current literature for modularity focuses mainly on techniques, tools, and on evaluation metrics. However, ontology developers still face difficulty in selecting the correct technique for specific applications and the current tools for modularity are not sufficient. These issues stem from a lack of theory about the modularisation process. To solve this problem, several researchers propose a framework for modularity, but alas, this has not been realised, up until now. In this article, we survey the existing literature to identify and populate dimensions of modules, experimentally evaluate and characterise 189 existing modules, and create a framework for modularity based on these results. The framework guides the ontology developer throughout the modularisation process. We evaluate the framework with a use-case for the Symptom ontology.