PhD of Islam Barchouch, INSA Renne, Defended on 3 December 2025,
Part of the ANR SKETCH project (grant number 21-CE38-0009).
Thesis supervisor: Eric Anquetil, Full Professor, INSA Rennes
Thesis co-supervisor: Nathalie Girard, Assistant professor, University of Rennes
https://hal.science/tel-05455772
This PhD thesis focuses on the conception of an Intelligent Tutoring System (STI), IntuiSketch, which supports learning by drawing using pen-based tablets. The system combines on-the-fly recognition of semi-structured sketches with real-time feedback generation to help students acquire scientific concepts through drawing. The application domain chosen during the thesis is anatomy sketches. The proposed architecture is based on two engines: a recognition engine, based on the CD-CMG formalism, extended by the use of fuzzy logic and a classifier, and a supervision engine based on a knowledge graph integrating the entities drawn by the teacher, the structural constraints connecting them and the planned educational sequence. The system operates in two modes: an author mode, enabling teachers to create exercises, and a student mode, enabling students to complete these exercises and receive real-time assistance.