Dealing with context-sensitive knowledge has been challenging in Artificial Intelligence in general and the Semantic Web in particular. The standard Semantic Web languages (RDF/RDFS/OWL) and reasoning tools do not explicitly consider the contextual dimension of knowledge, i.e., a particular statement (RDF triple) is valid only in certain circumstances known as contexts. With the rise of modern knowledge graphs like Wikidata (known as a property graph), the topic has gained increasing popularity. Property graphs are graph structures rich with contextual knowledge. For instance, Wikidata contains thousands of attribute-value pairs (i.e., qualifiers) to enrich its statements with information about time, space, causality, provenance, etc.

This thesis studies contextual knowledge representation and reasoning on 1) ontologies and 2) property graphs. First, in the context of ontologies, we present OWLC, an OWL extension for contextual reasoning. The target is to contextualize OWL axioms with instances and classes of contexts. Next, the applicability of this extension on modern knowledge graphs like Wikidata is examined. This study led to the second and core part of the thesis: reasoning on property graphs. This part starts by describing OWL-based languages’ limitations in the context of property graphs. Then, a logical framework that formalizes the impact of Wikidata qualifiers in reasoning is presented. More concretely, a contextual model that handles different categories of qualifiers is designed and formalized in a many-sorted logic. Each sort represents a qualifier category and is a high-level abstraction of several Wikidata qualifiers. Functions and axioms in a specific module of an algebraic specification are designed to define the test and combination operations specific to each qualifier category. Finally, we use this approach to formalize the rules corresponding to the ontological properties and some property constraints of Wikidata and present a corresponding rule-based reasoner with a prototype implementation.