Development of informative and telecommunication technologies have caused to create much dissimilar information. As well with growing different information resources in ontology designs, the importance of management these dissimilar resources has increased. In spite of most matchers use diverse measures for discovery the mappings, some semantic inconsistencies in final alignment are unavoidable. So it is essential to enhance a post-processing phase to training error patterns in the final alignment. The impartial of this research was refining the ontology semantic constraints over defining semantic constraints by a different measure for suitable weighting to the constraints. The outcomes indicated that the standard evaluation measures better in the suggestive method and comparing with other top ranked matchers the used method can create enhanced outcomes.