LPG-based Ontologies as Schemas for Graph DBs

Graph DBs are an emerging NoSQL technology that is boosting the opportunity of data handling based on interconnection and processing of single instances, rather than batch processing as usual in traditional relational DBs. Differently from relational DBs, the most prominent graph DB, Neo4j, is schema-less and based on the LPG graph model. We propose the definition and uses of ontologies as schemas, which would also enable high-level (logical) automated reasoning on the data. The graph model adopted by standard approaches to ontologies in Computer Science is incompatible with the LPG model. So, we propose a technology, called GraphBRAIN, specifically designed to exploit the full representational power of LPGs, still having a mapping to standard ontological approaches. GraphBRAIN also allows to apply different schemas on one underlying graph, representing different but inter-related views on the same data, and to combine schemas. This paper describes the formalism and outlines its possible applications. Development and implementation of the technology is ongoing, and a prototype is available and running.