Ontology matching algorithms for data model alignment in big data

Abstract

Big Data commonly refers to large data with different formats and sources. The problem of managing heterogeneity among varied information resources is increasing. For instance, how to handle variations in meaning or ambiguity in entity representation still remains a challenge. Ontologies can be used to overcome this heterogeneity. However, information cannot be processed across ontologies unless the correspondences among the elements are known. Ontology matching algorithms (systems) are thus needed to find the correspondences (alignments). Many ontology matching algorithms have been proposed in recent literature, but most of them do not consider data instances. The few that do consider data instances still face the big challenge of ensuring high accuracy when dealing with Big Data. This is because existing ontology matching algorithms only consider the problem of handling voluminous data, but do not incorporate techniques to deal with the problem of managing heterogeneity among varied information (i.e., different data formats and data sources). This research aims to develop robust and comprehensive ontology matching algorithms that can find high-quality correspondences between different ontologies while addressing the variety problem associated with Big Data.