Data Fusion in Ontology Based Data Integration

The existence of numerous data sources needs a common interface to access data for much research in the field of data analysis. Data Integration should address issues like semantic heterogeneity, redundant data and data conflict. Many approaches that were proposed lack in considering data conflict as an issue, thus the resolution results of them are often inaccurate. In this paper, a novel data conflict resolution approach based on Markov Logic Networks (MLNs) for Ontology based data integration is devised. Ontology is used to handle semantic heterogeneity that would result in increase in the recall rate and precision of the result is increased by resolving data conflicts and thus the system ensures that the user always receives high-quality data for the user query. Experiment on book dataset for online book purchase application is done and the precision of the resultant dataset is increased thus showing the effectiveness of this approach.