With the rise of information technologies and the Open Data movement, public data accessibility has significantly advanced, driven by initiatives aimed at increasing transparency and fostering innovation. However, despite these advancements, data interoperability challenges persist, primarily due to data heterogeneity. The Metropolis of Lyon, committed to an open data policy, has recognized these challenges, yet obstacles related to semantic heterogeneity limit the full integration and effective use of available territorial data.

In response to the challenges of data heterogeneity and interoperability, this CIFRE thesis explores solutions for the semantic enrichment of territorial data, with the goal of improving their interoperability and integration. The aim is to make these data both understandable and usable by human users and machines alike. In this context, the research draws on Semantic Web principles and introduces methods to support data producers, who are non-experts in the Semantic Web, in building Knowledge Graphs.

The contributions of this thesis include the design of a Semantic Model, called Annotated UML, intended for the semantic enrichment of data. This model forms the foundation for two proposed methods: the GES method and the sETL toolkit. These approaches facilitate semantic enrichment by converting data into RDF and integrating shared vocabularies without requiring extensive technical expertise in the Semantic Web. Specifically, the GES method involves data producers in the semantic modeling of their data, the creation of embedded documentation alongside the data, and the enrichment of data using external reference standards. The sETL toolkit leverages existing data schemas, that are the result of a partial elicitation of data semantics previously undertaken by the data producers of the Metropolis of Lyon. These methods aim to enhance open data governance and enable the integration of data into strategic and innovative initiatives for local governments.