Adapting a knowledge-based schema matching system for ontology mapping

In recent years, a large number of entities (ontology classes and properties) are found in different datasets over the Semantic Web. Due to the open and distributed nature of the Web, it is necessary to manage the heterogeneity problem between entities. In this context, the mapping of ontology entities from different datasets is important for data integration, data exchange and data warehousing. Existing semi-automatic ontology matching systems need some parameters such as thresholds and weights, and send the results to the users for adding correct and removing incorrect mapping manually. However, there is no existing solution for correcting these mappings automatically. The main goal of our research work is to do ontology mapping by adapting our Knowledge-based Schema Matching System (KSMS) that allows users to correct and validate the matching results automatically. Our approach is based on Hybrid Ripple-Down Rules (RDR) that combines machine learning and knowledge acquisition approaches. In the hybrid approach, first a machine learning algorithm is used for classifying entities, and then rules are added by incremental knowledge acquisition for solving matching errors such as false positives and false negatives at the element level. The system also computes structure level matching considering hierarchical structure of a full graph. In this research, we perform experiments on the conference track of the ontology alignment contest OAEI 2014. Experimental results demonstrate that our system improves performance in terms of precision, recall and F-measure.