A multi-attribute based framework for ontology aligning

摘要

This paper is a comprehensive description of our proposal of ontology alignment framework and novel evaluation procedure. The main contribution is providing precise definitions of ontologies on every level of their granularity. On top of such foundations we have built a set of consistent algorithms that allow designating clear matches between two ontologies. The difference between our solution and former approaches to this task is in accepting concepts' attributes and their varying semantics as the grounding level of every workflow that has been defined. Based on the conducted experiment, we can claim that such an approach is useful and reliable. Due to the fact that created framework is the complete redefinition of the problem, we could not incorporate widely used experimental procedure developed by Ontology Alignment Evaluation Initiative. This is the reason why we have created a novel approach to alignment evaluation based on statistical SignTest that can be conveniently used to emphasise differences in quality of designated mappings.