Towards a framework for tensor ontologies over Neo4j: Representations and operations

Abstract

Ontology has been an active research field connecting philosophy, logic, history, mathematics, and computer science to name a few. Within an ontological context defined over a domain the entities as well as their associated relationships can be represented by the vertices and the edges of a tree. From the latter new knowledge can be then inferred through a number of techniques including Horn logic from reasoners and RDF triplets. With the advent of the Semantic Web and sophisticated associated software tools including graph databases such as Neo4j, Sparksee, and TitanDB or XML parsers such as Xerces graph mining is done efficiently on the semantic level instead of the combinatorial or algebraic ones. Multilayer graphs, namely graphs whose labeled edges belong to a number of predetermined classes, have been recently introduced in social network analysis in order to represent the different interaction options between netizens. In this work, the potential of applying this new type of graphs to an ontological context creating essentially an ontological tensor is outlined and its complexity is assessed. A human readable dataset based on the late 1970s and early 1980s Apple manually constructed from the 2011 officially authorized biography of Steve Jobs and the 1999 film Pirates of Silicon Valley serves as a concrete example complete with Neo4j queries.