Technical Report. September 2016. NdFluents: A Multi-dimensional Contexts Ontology Jos´ e M. GIM ´ ENEZ-GARC ´ IA a , Antoine ZIMMERMANN b and Pierre MARET a a Universit´ e de Lyon, CNRS, UMR 5516, Laboratoire Hubert-Curien, Saint- ´ Etienne, France b ´ Ecole Nationale Sup´ erieure des Mines, FAYOL-ENSMSE, Laboratoire Hubert Curien, F-42023 Saint- ´ Etienne, France Abstract. Annotating semantic data with metadata is becoming more and more important to provide information about the statements being asserted. While ini- tial solutions proposed a data model to represent a specific dimension of meta- information (such as time or provenance), the need for a general annotation frame- work which allows representing different context dimensions is needed. In this pa- per, we extend the 4dFluents ontology by Welty and Fikes—on associating tem- poral validity to statements—to any dimension of context, and discuss possible is- sues that multidimensional context representations have to face and how we address them. Keywords. Ontologies, OWL, Context, Reification 1. Introduction In knowledge representation, it is often necessary to characterize the context associated to a statement, such as when and how it was generated, or who uttered it. However, RDF and OWL can only represent natively binary relations [12]. Many models exist for pro- viding statements about statements, some of which are specific to a certain “dimension” of context, such as temporal validity. Along these lines, in 2006, Welty and Fikes [17] proposed an ontology for describ- ing fluents (i.e., entities whose characteristics change over time). Their approach ad- vanced the state of the art in temporal representation, and has been used and extended in other works, but it only addresses one dimension characterizing a statement. Nonethe- less, Welty and Fikes’s approach can be extended so that any number of context’s dimen- sions can be represented. We propose this extension through a generic ontology that can be extended to implement any specific dimension of context. In addition, we address the problem of modeling contextual datatype properties, as well as combining different con- textual dimensions. This work is motivated by the need to characterize web datasets in terms of trust, provenance, and temporal validity, in the context of a question answering system within the WDAqua project. 1 1 http://wdaqua.informatik.uni-bonn.de