Mediating Schema Ontology for Linked Open Data Cloud Integration Using Bottom-Up Schema Mapping

Abstract

Flexible data integration from heterogeneous sources, formalization of database schema, and consequent support and facilitation of knowledge discovery have become crucial to gaining competitive advantage. Consequently, its validity ("building the right system"), verification ("building the system right"), and efficiency are key factors, which enable more effective intelligent mining and recommendations. The Linked Open Data Cloud (LoD Cloud) is Cloud-based open data, which can be accessed using open standard protocols, such as HTTP, URL and URI. Enabling automated ontological information integration has commonly involved ontology as a mediator, which reconciles and provides a unified view of data structure, and as a mapping interface with the source database schema. Our approach to ontology versioning creates a mediating ontological layer, which encompasses a set of ontological properties which become updated based on queries by the user. We extend our prior work in terms of the change of data type property, object property and IS-A relationship are included in this paper and method/behavioral changes. The experimental results indicate 87% Precision, 100% Recall and 93% F-measure. These indicate that our approach has successfully captured changes in the database schema as changes in mediated ontology. We have also addressed the problem of change alignment when overlapping topics are discovered in mapping ontologies via formal validation and optimized via description logics consistency checking.