We offer a PhD position in the field of computational semantics, entitled "Predicting discourse moves". Supervisors : Marta Abrusan Nicholas Asher , IRIT group : MELODI Description: Interaction with automated dialogue systems is ever more present in every day life. Yet, our understanding of the theoretical and computational aspects of interactive dialogues is still limited. We propose a project in this domain. The project is based on the Settlers corpus, which was conceived as a tool for studying the nature of strategic conversation within the STAC project (http://www.irit.fr/STAC/index.html). Settlers of Catan is a win-lose game in which players use resources, e.g. wood, ore and sheep, to build roads and settlements. In the course of the game the players negotiate for these resources. Our corpus consists of the records of these negotiation dialogues (over 900 dialogues), as well as the records of the trades and computer-dealt resources. The corpus records both the linguistic and the non-linguistic environment in which dialogue moves are made. The records of the negotiation dialogues have been annotated for their rhetorical structure using the rhetorical model of Segmented Discourse Representation Theory (SDRT, Asher and Lascarides 2003). Machine learning experiments have been also performed to automatically infer the rhetorical structure of dialogues. The thesis project we propose draws on the results in STAC and aims at predicting the next move in a Settlers game based on the linguistic and the non-linguistic context. It has both a computational and a theoretical aspects. The computational aspect draws on the results of the machine learning experiments to predict the type of the most coherent follow-up discourse move. The theoretical aspects include (a) the understanding and calculation of strategic interests of conversational participants in a game theoretical setting using models that have been developed to deal with non-cooperative settings (b) further development of the formal rules of discourse coherence in a conversation, taking into account presuppositions, topicality and focus structure (c) the theoretical integration of the linguistic and non-linguistic information in a dialogue. For more information, contact Marta Abrusan or Nicholas Asher .