PhD subjects

Printer-friendly versionPDF version

Below are some research interests that could be explored as part of a PhD research project.
Be aware however that these are non funded positions: you have to provide your own funding.
Personality for companion agents
Conversational agents have now been endowed with emotions for a few years, making them more realistic during their interactions with users. What they are now lacking is to have a consistent personality that would differentiate agents from each other, and that would make them engaging in the long-term. Things to be investigated in particular are the links between personality and emotions, between personality and interactions, and how to choose a personality for the companion that best fits its user's personality in order to facilitate their interactions. Possible applications include companion agents, or video games.
Trust and interaction
This topic is about investigating the role of trust in interactions between agents. It is usually assumed in agent communication languages (ACL) that agents trusts each other, but this should not always be the case. The goal here is thus to provide a novel semantics for ACL where trust between the partners is explicitly taken into account, instead of being implicitly assumed to exist. This will lead to different interaction protocols between agents depending on their level of trust in their partner. Questions to be solved in particular are the influence of trust on the selection of an appropriate protocol, and the evolution of trust over time as a result of past interactions. Possible applications include business to business automated exchanges.
Coping with emotions
Emotional artificial agents are getting more and more common. However the focus is usually on agents able to express realistic enough emotions for the sake of engaging interaction with human users. On the contrary this topic is centered on a model of emotions that is faithful to psychology; the goal is more to accurately simulate a human. With this goal in mind, we would like to go further than the models of emotional appraisal commonly used in computer science, and also consider and formalise the notion of coping. Coping is a psychological process by which an individual adapts to his emotion and tries to stay or come back to a state of equilibrium (for instance deny a bad news, or change your mind from it). Concretely, this work will start from an existing BDI logic model of the appraisal process and extend it to account for the coping process. Expected difficulties are to deal with its dynamic and temporal aspects. Possible applications are human behaviour models for a variety of agent-based social simulations.
Memory and emotions
This project is a bit similar to the previous one in that it also consists in trying to accurately simulate human behaviour, so an interest in psychology is needed. The focus here is on memory, and in particular simulating how humans remember pieces of information in relation to their current emotion. Psychological theories will be used to determine the role of emotions in storing, retrieving, deleting chunks of memory. This will then be formalised and reproduced in an artificial agent. Possible applications also concern agent-based social simulations.