Enhancing CIDOC-CRM Models for GeoSPARQL Processing with MapReduce

Spatial and temporal dimensions are two important characteristics of archaeological data and cultural heritage in general. The ability to perform some sort of reasoning on them is crucial during the analysis and interpretation process performed by domain experts. Many models have been defined in literature in order to properly describe such data and support the following interpretation process; among them, CIDOC CRM is a formal ontology specifically developed to represent cultural heritage information and many extensions have been proposed in recent years in order to enrich such model. In particular, CRMgeo tries to bring the gap between the cultural heritage domain and the geo-spatial domain, by providing a link towards GeoSPARQL and by defining the necessary constructs for the representation of spatial data types and relations. Unfortunately, the current support to the process of spatial functions through SPARQL query engine is still limited and many performance problems remain. The aim of this paper is twofold: (i) to evaluate the applicability of CRMgeo in representing spatial characteristics and relations of archaeological objects, and (2) to propose a MapReduce procedure able to efficiently derive spatial relations between objects, in order to automatically enhance an RDF model with them and avoid the performance issues derived from the use of GeoSPARQL query engine.