The notion of territory plays a major role in human and social sciences. Representing this spatio-temporal object and computing their changes have been tackled in various ways. However, in a historical context, most existing approaches are irrelevant as they rely on geometric data, which is not always available. Thus we designed the HHT ontology (Hierarchical Historical Territory) to represent hierarchical historical territorial divisions, without having to know their geometry. Previous versions of the ontology were limited in this aspect by the notion of building Block. This notion was expanded to enable a more flexible representation of geometry, notably based on set operators. An algorithm to detect and qualify changes was implemented using the new geometry formalisation. Six knowledge graphs regarding the evolution of French communes and the expansion of New York City were created to evaluate the use of the ontology and the proposed algorithm.

The development of the Semantic Web of Things (SWoT) is challenged by the nature of IoT deployment architectures, where constrained devices collect data processed remotely by powerful Cloud servers. Such a deployment pattern introduces bottlenecks constituting a hurdle for scalability, and increases response time. This hinders the development of a number of critical and time-sensitive applications. Enabling the deployment of the Semantic Web stack closer to the constrained devices of the IoT may foster the development of time-sensitive interoperable applications, while reducing forwarding the user data to remote third party Cloud servers. The approach we develop in this paper is a contribution towards this direction, and aims to enable rule-based reasoning closer to sensors producing IoT data. For this purpose, we define a distributed scalable semantic processing algorithm by dynamically propagating deduction rules on Fog nodes. Our goal is to shorten the time needed to deliver high level information deduced from the collected data. This approach is evaluated on a smart building use case where both distribution and scalability have been considered.

In this paper we present a system that allows a user to explore or mine a document collection. This system is based on domain and task knowledge modelled in the form of ontologies and allows direct access both to information as it is stored and to information that is built from it. The system has been developed in Java.