Semantic Layer Construction for Big Data Integration

Heterogeneity problem (data model, schema and semantics) is the one of the challenges of data integration from different big data stores. In order to overcome that problem in big data integration, semantic conceptual layer is constructed as an intermediate layer between different data stores by using ontology. To achieve this goal, there are two steps involved: generate local ontologies from different data systems and merge extracted local ontologies to build a global ontology. The main focus of the paper is merging ontologies which matches local ontologies by syntactic and semantic similarity measures. The two concepts are syntactically compared by Jaccard similarity measure and semantically compared by using WordNet. The matching approach takes into account the class name, internal structure (attributes) of the class, relationship (objectProperty relation, is-part-of relation) of the compared concepts. The performance of proposed algorithm will be compared other merging algorithms such as PROMPT.