[agents] PhD Vacanacy Self-Explaining Cognitive Agents

[Koen Hindriks - EWI]

We are looking for excellent candidates that have received an MSc degree in Artificial Intelligence or Computing Science. Candidates that have experience with cognitive agent technology or more specifically with agent-oriented programming, and/or automated explanation are invited to apply for this position. Candidates should have excellent Java programming skills and preferably a strong background in software engineering. Ideally, the candidate also has experience with user evaluation of software tools.

For further details on this position, please contact Dr. K.V. Hindriks (K.V.Hindriks at tudelft.nl<mailto:K.V.Hindriks at tudelft.nl>). Please see www.ii.tudelft.nl<http://www.ii.tudelft.nl> on how to apply.

Project description
Cognitive agents derive their choice of action from their beliefs and goals. This potentially provides these agents with the capability to self-explain their behaviour in terms of these concepts. Applications that are built using cognitive agents that can self-explain their behaviour can provide users with a better understanding of the application. A recommender agent developed using cognitive agent technology could, for example, explain the recommendations it makes to its user. An application domain that will be used in the project is that of a negotiating agent that is able to explain why it offers the bids that it does in a negotiation. Agents that can explain their choice of action are useful not only for end users but also for the developers of such agents. An agent developer needs to identify defects in the agent programs s/he writes. Debugging an agent program requires analysing the behaviour generated by a cognitive agent program that is unexpected or undesired. In order to understand how this behaviour is generated, it would be of great value to a programmer if an agent can explain why it chose a certain behaviour.

In order to design and develop self-explaining cognitive agent technology, the candidate will work on mechanisms that can derive explanations from the traces generated by a cognitive agent program. One challenge is to efficiently keep track of the history of a cognitive agent's mental states and choices and to design the reasoning technology that can derive explanations from this data and the agent program itself. Enabling agents to become more self-explanatory requires work on a language for explaining behaviour and associated reasoning technology as well as on tools that allow a user to trace the behaviour of an agent and to generate explanations for specific behaviour episodes. The aim is to apply techniques from the area of omniscient debugging and develop techniques that can be integrated into logic-based agent programming languages. Another challenge is to show that the explanation capability developed facilitates both programmers in debugging as well as users in understanding agent behaviour. We therefore aim to perform experimental studies to evaluate how programmers can benefit from self-explaining agent technology and how a user gains an improved understanding of an agent in a negotiation scenario.

The PhD will be part of the Interactive Intelligence section. The PhD position is part of a project on agent-oriented programming that aims to design and develop cognitive agent technology that supports and simplifies the engineering of agent-based systems. For more information on the Interactive Intelligence group, please visit www.ii.tudelft.nl<http://www.ii.tudelft.nl>.