A data model-independent approach to big research data integration. (30th September 2019)

• Record Type: Journal Article
• Authors: Bartalesi, Valentina; Meghini, Carlo; Thanos, Costantino
• Is Part Of: International journal of metadata, semantics and ontologies. Volume 13:Number 4(2019)
• Page Start: 330
• Page End: 345
• Publication Date: 2019-09-30
• Subjects: data integration -- big research data -- ontology -- semantic web; Metadata -- Periodicals; Semantic Web -- Periodicals; Ontologies (Information retrieval) -- Periodicals; Data structures (Computer science) -- Periodicals; Information theory -- Periodicals; 005.74
• Languages: English
• ISSNs: 1744-2621

Abstract

The paper addresses the data integration problem in the context of the scientific domain. The main characteristics of the big research data that make the traditional approach of data integration unfeasible are presented. Two new emerging practices, i.e. an exploratory approach to data seeking and an empiricist epistemological approach to knowledge creation, are discussed. Based on these considerations, we present a new paradigm of data integration and an application ontology that supports it. The ontology is based on five types of events and every event is extensionally modelled as an input/output operation on the involved data entity. The strong point of the ontology and of the whole approach to data integration is that no assumption is made on the data models in which the databases or the views are expressed. This provides a level of generality that successfully deals with the heterogeneity of the domain.