A New Approach Based on the Bee Optimization Algorithm for Ontology Alignment: ABCMap+

Background

Ontology alignment allows to generate matches “similarity or dissimilarity” between the entities. These entities are concepts or properties or even instances. However, the problem is that the parameterization in the ontology alignment systems is essential since it is possible to significantly modify the similarity values. The modification of some parameters can be done manually by a human expert familiar with the context. This task imposes perfect control of the alignment system. However, such a choice of the parameter is neither always desirable nor possible for large ontologies. These include for example, when the similarities between concepts of the source ontologies are calculated, aggregation is performed to find similarities between the combined concepts.

Since the proposed alignment system in this paper, called ABCMap+, exploits many similarity calculation factors, aggregation is necessary to find the best possible mapping candidates. In order to overcome this problem, based on current development of ABC (Wang and Zhang, 2016) (Tang et al., 2017), we propose an artificial bee colony-based optimization algorithm to generate similarity aggregation based on optimal weights in order to obtain a well-optimized alignment.