PhD student specialising in the detection and semantic annotation of changes in geographical data -LIG

Descrizione

PhD student specialising in the detection and semantic annotation of changes in geographical data -LIG Sintesi in italiano (traduzione automatica): L'organizzazione internazionale cerca un dottorando specializzato nella rilevazione e annotazione semantica dei cambiamenti nei dati geografici. Il progetto, parte dell'ANR GEvoK, si concentra sull'analisi delle evoluzioni delle banche dati geografiche come OpenStreetMap. Il candidato dovrà studiare le strutture dati di OSM e del database Topo, progettare un approccio per la rilevazione automatica dei cambiamenti utilizzando tecniche di intelligenza artificiale e implementare un modello semantico per descrivere e annotare i cambiamenti rilevati. È richiesta una laurea in discipline affini, come Geografia, Informatica o Ingegneria, e competenze in QGIS, Python, machine learning e tecnologie del web semantico. Geographic databases such as OpenStreetMap (OSM) and the IGN Topo database are constantly evolving to reflect changes in the landscape. Identifying, describing, and characterizing the changes these geographic databases undergo over time is a major challenge for analyzing how landscapes evolve over time. This thesis is part of the ANR GEvoK (Geographic Entities Evolution in Knowledge Graphs) project, which aims at automatically detecting and semantically annotating geographic changes in data from various sources (satellite or collaborative sources such as OpenStreetMap, for example). The approach combines artificial intelligence (AI) techniques for the automatic detection of changes over time, semantic web technologies for the representation and enrichment of knowledge derived from this detection, and finally LLMs for querying the resulting knowledge bases and generatingnarratives describing the changes in natural language. The PhD student’s main responsibilities will include: - studying the data structures of OSM and the Topo database (models, granularity, semantics, attributes) using QGIS and Python for data exploration and preprocessing - designing an approach for automatic change detection using AI techniques (classification, clustering, or supervised/unsupervised learning depending on the available data) using spatial analysis and geographic machine learning tools (e.g., scikit-learn, PyTorch/TF + GeoPandas/Shapely) - Implementing a semantic model (ontology) to describe and annotate detected changes using the Semantic Web (RDF, OWL, SPARQL) - Integrate the results into a geographic knowledge graph (KG) - Enriching the KG by integrating contextual elements from the Web or LLMs to better explain and validate the detected changes (causes, pressures, dates, geopolitical context, etc.) - Publishing and disseminating the KG to a wide audience (experts, researchers, citizens) via natural language query tools based on LLMs (avoiding the use of SPARQL) and metrics to analyze changes Annuncio in inglese. Fonte: Euraxess (Commissione europea).