Exploiting data dependencies with null values for ontology extraction

Abstract

I present a novel methodology to automatically extract ontologies in the form of conceptual schemes characterizing raw data and then generate mappings linking the obtained ontology to raw data with the aim to extend Ontology Based Data Access to raw data. Additionally, I propose new definitions of functional and multivalued dependencies under SQL (in short, resp., nFDs and nMVDs) to model the semantics of raw data and to achieve raw data normalization. The main novelty in this work is to adapt existing techniques to deal with the case of data with missing or incomplete information that is witnessed through the presence of null values. © 2012 Springer-Verlag.