Abstract

The ACACIA project is a multidisciplinary project that aims at offering models, methods and tools in order to help the knowledge engineer to acquire (i.e. elicit and model) knowledge from multiple expertise sources (experts or documents). The purpose is to capitalize expertise on a given field or to build a knowledge-based system.

Research Actions

For knowledge acquisition from multiple experts, we defined a protocol for elicitation of expertise from a group of experts. We proposed a model of cognitive agent for guiding knowledge acquisition and we proposed several techniques of comparison of multiple expertise models: comparison of knowledge graphs, generation of consensus rules. We also developed an architecture of cognitive agent and a platform for building multi-agent applications.

We analysed explanatory dialogues among experts solving cooperatively a problem and we determined the explanatory strategies appearing in such dialogues. We improved our explanatory text planner developed last year. We studied a modelling of evaluation activity.

We proposed techniques for analysing expertise documents, in order to model the underlying expertise or explanatory knowledge. We also developed a tool allowing to build conceptual graphs from electronic documents and managing hypertext links between elements of documents and the conceptual graphs of a base.

We proposed several methods for acquiring qualitative knowledge such as topoï and we improved our qualitative reasoning engine for performing simulations on knowledge graphs, in order to validate the acquired knowledge.

Industrial and International Partnerships

The project is funded by:
[English Abstract]
[Table des Matières]  [Index des mots-clés]  [Index des personnes]