Skip to Main content Skip to Navigation

Dialoguer pour décider : Recommandation experte proactive et prise de décision multi-agents équitable

Fabien Delecroix 1 
1 SMAC - Systèmes Multi-Agents et Comportements
CRIStAL - Centre de Recherche en Informatique, Signal et Automatique de Lille - UMR 9189
Abstract : If decision making can be a pure individual process, it can involve several actors and present social aspects. In this thesis, I consider two types of social decision process : supported decision making and collective decision making. Concerning supported decision making, two actors have distinct roles : the decision maker and the assistant. Here, the decision maker is a human agent and the assistant a software one. In many applications, the dialogical abilities of the assistant are deceptive and the dialogue lacks of consistency. To tackle this problem, we design a proactive dialogical agent aiming for the credibility in conversation and the relevance of recommendations : our agent leads the conversation in asking relevant questions to collect the preferences of the decision maker and use them in recommending the alternatives that fit the most. We apply our approach on the e-commerce field. The second contribution concerns collective decision. The objective is to define a process that lead to a fair agreement, even if participants have incomplete preferences. For this purpose, I define the fair agreements by applying the leximax criterion on the rank of alternatives. Then, I propose a negotiation protocol to reach such agreements and the strategy is taken into account to evaluate it. Finally, the protocol is applied to the search of a meeting point in a maze.
Complete list of metadata
Contributor : CRIStAL Equipe SMAC Connect in order to contact the contributor
Submitted on : Friday, December 18, 2015 - 6:01:02 PM
Last modification on : Wednesday, March 23, 2022 - 3:51:03 PM
Long-term archiving on: : Saturday, April 29, 2017 - 10:18:04 PM


  • HAL Id : tel-01246616, version 1


Fabien Delecroix. Dialoguer pour décider : Recommandation experte proactive et prise de décision multi-agents équitable. Système multi-agents [cs.MA]. Université Lille1 - Sciences et Technologies 2015. Français. ⟨tel-01246616⟩



Record views


Files downloads