The Effect of Semantic Constraints on Ontology Alignments using Neural Network

Improvement of informative and telecommunication technologies have caused to create much heterogeneous information. In addition with increasing various information resources in ontology formats, the importance of managing these heterogeneous resources is increased. Despite most matchers use different measures for finding the mappings, some semantic inconsistencies in final alignment are inevitable. Therefore it is necessary to add a postprocessing step to study error patterns in the final alignment. The objective of this research is two basic improvements of the ontology semantic constraints with defining semantic constraints and using a new measure for appropriate weighting to the constraints. Evaluation results show the standard evaluation measures are improved in our suggestive approach. Comapring with other top ranked matchers our approach can produce better results.