Concept type and relationship type classification based approach for identifying and prioritizing potentially interesting concepts in ontology matching
S. Ranjini, K. Saruladha · 2016
Due to large and heterogeneous data present across various ontologies which are developed by different knowledge engineers with various backgrounds describe the concepts and their relations using different terminologies has lead to the construction of several ontologies with different or same terminologies for similar domain. The heterogeneity among different ontologies for representing the similar domains limits interoperability across the ontologies. For effective utilization of ontologies and to solve the heterogeneity problems ontology matching is used which is a technique that determines the matches between the concepts that are associated in distinct ontologies developed for the same domain. Due to the growth of large number of ontologies and increase in the size of the ontologies rapidly the need for search space optimization also becomes a challenging task. This can be handled by potentially identifying and prioritizing the rich semantic concepts and their associated relations from the large ontologies thereby filtering out the less important concepts using the proposed concept type and relationship type classification method by assigning weights to different concepts and their relations experimentally. This proposed methods takes into consideration of the existing concept type classification approach[1]and classifies into five novel concept types by assigning different weights to concepts that comes under different classifications and also proposes a method to assign weights for different types of relationships associated between two or more concepts. The proposed approach is different from the existing ranking and concept importance methods which assigns weights to different concepts and its associated relations based on certain features iteratively. The proposed approach is proved to be significant by experimentally assigning weights to a relatively small set of clusters of an ontology and the results helps to reduce the search space optimization of the ontology there by increasing the efficiency and effectiveness of the ontology matching system.