CartoAI-Conflict: Multimodal and Explainable AI for Cartographic Conflict Detection and Map Generalization Support
[CartoAI-Conflict: multimodalne i wyjaśniane modele AI w identyfikacji konfliktów kartograficznych i wspomaganiu generalizacji map (01.08.2026-31.12.2026)].Key words
GeoAI; multimodal AI models; explainable AI; human-in-the-loop; cartographic generalization kartograficzna; cartographic conflicts
Project summary
Automatic generalization still requires expert decision-making, particularly in the identification of spatial conflicts, assessment of their significance, and selection of appropriate generalization operations. Existing AI-based methods focus primarily on object selection, classification, and geometric transformation, while research on multimodal models has largely addressed map interpretation, question answering, georeferencing, and map style. Systematic research integrating conflict detection, object identification, operation selection, explanation evaluation, and the incorporation of expert preferences is still lacking. The novelty of the proposed approach lies in treating the model as an explainable diagnostic and decision-support system and in integrating formalized cartographic knowledge, explainable AI (XAI), and a human-in-the-loop approach. The results will demonstrate which generalization decisions can be effectively supported through automation and which should remain under the cartographer’s control.
Finansed by: Excellence Iinitative Research University (IDUB).
Team
- Izabela Karsznia (Principal Investigator)
- Iga Ajdacka (Doctoral researcher – employed in the project)
- Jakub Frommholz (Student researcher – scholarship holder)
This research was funded by the University of Warsaw, Poland, within the initiative „Research Impulses”, Excellence Iinitative Research University (IDUB).
