A Discovery is, in general, the action of finding what was not known before. In science, we speak obviously of Scientific Discovery. The field of Scientific Discovery in Artificial Intelligence (AI) focuses on the development of integrated and autonomous systems for scientific discovery. Several sub-fields of AI study problems that can be linked to the broader issue of scientific discovery. This is the case, for example, of expert systems and data mining systems allowing to build discovery programs using the techniques of problem solving.
In this context, several families of discovery systems have emerged since the birth of the field of scientific discovery in AI. In this thesis, we built a system based on data from actual experiments done by Claude Bernard. This system, called CYBERNARD, reconstructs rationally the empirical method of the latter by modeling his experiments based on actual results. In the case of CYBERNARD, the role of the machine is to simulate the experiments performed by Claude Bernard to get virtual experiments on the system's output. The system is adjusted until the virtual experiments are consistent with those made by Claude Bernard himself in order to test the hypotheses inferred.
The construction of the system CYBERNARD has started with an epistemological study of Claude Bernard's manuscripts until building a computer model simulating a part of his discovery process. During the epistemological study, we studied the several modes of inference used in Claude Bernard's experimental process and represented the knowledge contained in his manuscripts (ontology, experiments) to be able to facilitate the use of this knowledge in our computer model: the virtual laboratory. This virtual laboratory is the environment in which the simulation of Claude Bernard's experiments is proposed to the user. The purpose of this laboratory is to make virtual experiments and to compare them with actual experiments made by Claude Bernard. We propose a virtual laboratory that contains core models (virtual organisms built on the hypotheses and on which Claude Bernard's experiments are done) and parameterized operators (operators from the descriptions of experiments included in Claude Bernard's writing). Finally, we validate our approach on scenarios drawn from some of Claude Bernard's experiments.
Defesas : 26/11/2010 - 14h - Site Jussieu 55-65/211
Membros da banca :
Pr. Claude Debru - ENS
Pr. Jean-Gabriel Ganascia - LIP6
Pr. Jean-Marc Labat - LIP6
Pr. Lorenzo Magnani - Univ. Pavia [Rapporteur]
Pr. Jean Sallantin - LIRMM [Rapporteur]
- Ch. Jouis, C. Jouis, F. Guy, B. Habib, J.‑G. Ganascia : “Combination of Topology and non-monotonic Logics for Typicality in a Scientific Field: Paleoanthropology”, Twenty-Fourth International FLAIRS Conference, Palm Beach, Florida, United States, pp. 174-179, (AAAI) (2011)
- B. Habib : “CYBERNARD : Un Système de Découverte Autonome qui Reconstruit Rationellement la Démarche Empirique de Claude Bernard en Modélisant ses Expérimentations Basées sur des Résultats Réels”, tese, defesas 26/11/2010, direção de pesquisa Ganascia, Jean-Gabriel (2010)
- Ch. Jouis, F. Guy, B. Habib, J.‑G. Ganascia : “Le problème de l’émergence dans les textes scientifiques ou techniques : modélisation ontologique et application à la paléontologie”, ACFAS, Québec, Canada, (ACFAS (Association Francophone pour le savoir)) (2010)
- Ch. Jouis, B. Habib, J.‑G. Ganascia : “Atypicalities in Ontologies: Inferring New Facts from Topological Axioms”, ARCOE 2009, “Automated Reasoning about Context and Ontology Evolution”, Pasadena, California, United States, pp. 43-45 (2009)
- C. Laudy, B. Habib, J.‑G. Ganascia : “Fusion of Claude Bernard’s Experiments for Scientific Discovery Reasoning”, 17th International Conference on Conceptual Structures, ICCS 2009, vol. 5662, Lecture Notes in Computer Science, Moscow, Russian Federation, pp. 219-232, (Springer) (2009)
- B. Habib, J.‑G. Ganascia : “The Reasoning Process underlying Claude Bernard’s Scientific Discoveries”, IJCAI WORKSHOP ON GRAPH STRUCTURES FOR KNOWLEDGE REPRESENTATION AND REASONING (GKR 2009), Pasadena, California, United States (2009)
- B. Habib, C. Laudy, J.‑G. Ganascia : “Using Fusion to Fill in the Gaps in Old Scientific Discoveries’ Notebooks”, IJCAI WORKSHOP ON GRAPH STRUCTURES FOR KNOWLEDGE REPRESENTATION AND REASONING (GKR 2009), Pasadena, California, United States (2009)
- Ch. Jouis, B. Habib : “Exceptions in Ontologies: Deducing Properties from Topological Axioms”, Proceedings of the 22nd International Florida Artificial Intelligence Research Society Conference (FLAIRS 22), Sanibel Island, Florida, United States, pp. 512-513, (AAAI) (2009)
- Ch. Jouis, J. Bourdaillet, B. Habib, J.‑G. Ganascia : “Exceptions in Ontologies: A Theoretical Model for Deducing Properties from Topological Axioms”, chapter in Ontology Theory, Management and Design: Advanced Tools and Models, (IGI-Global) (2009)