Proposal For Master Project : Learning Conversation

The goal of the project is to propose and implement a knowledge representation system suited for conversation theory.
Entailment meshes, a model proposed by the conversation theory authors (G. Pask, Paul Pangaro) will be first considered, and then compared to more recent approaches.

The strength of entailment meshes is their coherence driven efficient dynamic knowledge management. Meaning conflicts happening in conversation situations trigger the creation of new «bubbles» enabling to deal with new possibilities of meaning, while the entire system is always tending to coherent states.

The chosen model will be implemented and tested on a prototypical situation of human-machine conversation, e.g. in the context of a small application of adaptive coach in conversation with a human, or of smart environment management.

Bachelor works contributing to the project, as well as PhD Research works with extended purposes, are also possible on demand

Keywords: Human-machine conversation, conversation theory, entailment meshes, cybernetics, learning conversation, Agile Interactive Systems.

Contact: Dr Michèle Courant