Nos partenaires

CNRS

Rechercher





Accueil du site > English > Research Topics > Topic 4 - Reasoning and Decision

Topic 4 - Reasoning and Decision

Head : Andreas Herzig

 

 


The "Reasoning and Decision" theme investigates modelling and automating processes of reasoning and decision making. Its research topics are situated in the areas known as artificial intelligence and computational linguistics and aim at tackling the following issues :

  • Modelling and reasoning about beliefs and their dynamics : logical models for the modelling of incomplete, uncertain and/or partially incoherent beliefs ; belief revision ; reasoning about actions and about causality ; belief fusion ; machine learning. The models that we use include various theories of uncertainty (probabilities, belief functions, possibilities, purely ordinal models) and various non-classical logics (epistemic and doxastic logics, dynamic logic, possibilistic logic and other logics of uncertainty, paraconsistent logics).
  • Decision aids and automated decision : decision under uncertainty, multicriteria or group decision, sequential decision, and planning. Our research aims at normative models, preference representation languages, and on the development of efficient algorithms (in particular for constraint satisfaction and planning).
  • Formalisation of interaction and communication between agents : dynamic theories of discourse and dialogue ; formalisation of speech acts, agents’ intentions, and argumentation and negotiation processes between agents ; game theory and logical aspects of strategic interaction.
  • Security of information and communication systems : verifying cryptographic protocols and deciding consistency of security policies.
  • Models and methods for natural language processing, with a focus on the semantic and pragmatic dimensions of natural language : formal semantics, lexical semantics, semantics-pragmatics interface in discourse ; statistical methods and learning from corpora ; applications to opinion and preference analysis, information extraction, the analysis of the discursive structure of texts and the analysis of non-cooperative dialogues.
  • Knowledge representation by means of ontologies : we study the theoretical foundations and the way the formalisms (such as description logics and also graphs) and the fundamental knowledge organisation principles allow to account for a precise semantics ; we also study the questions coming with their construction, their interplay with natural language and lexical semantics, their querying, their alignment, and their use for information retrieval and reasoning.

The "Reasoning and Decision" theme is made up of three teams : ADRIA, LILaC and MELODI. The interaction between these teams is important and comprises co-supervised PhD theses and regular seminars of the theme. The activities of the theme increasingly involve interdisciplinary collaborations with researchers in linguisitics, cognitive psychology, philosophy and economy. The teams of theme 4 also interact with IRIT’s other themes and contribute to transverse projects.

Cette rubrique ne contient aucun article.