6 - 8 weeks
Confidence with programming in Java, a basic understanding of Prolog, acquaintance with Intelligent Agents and AI concepts, as well as eagerness to get familiar with automated negotiation tools and techniques.
We negotiate every day in our life for a number of activities ranging from deciding the wake-up time in the morning, the number of hours to spend on our coursework, a destination for our next holiday to the number of bucks to be spent on our desirable gadgets. How great it would be if there is an agent to negotiate on behalf of us, and decide the best bargain with its decision-making capability. To support this idea, one approach that has been proposed is to build software agents that behave like automated negotiators on behalf of users.
In this project, the UROP recipient will get acquainted with a General Environment for Negotiation with Intelligent multi-purpose Usage Simulation (GENIUS) (see ) to develop a negotiating agent. In this context, the student will be given a heuristic negotiation strategy called CONAN (already implemented for a single-issue domain using Prolog in RECON, see ). This agent strategy will be developed on GENIUS (a Java-based negotiation platform) using a very simple negotiation protocol for a multi-issue negotiation domain in an e-marketplace. The negotiation platform will allow the student to compete his/her agent against the already existing agents in GENIUS and evaluate them based on a number of parameters, for instance, agent utility.
As the proposal results from an existing research collaboration between Pallavi Bagga (PhD Candidate), Dr Nicola Paoletti and Prof Kostas Stathis, the successful candidate will need to participate in group meetings and be co-supervised by the current members of the group.
We anticipate that this project will help the student assess his/her programming skills, and develop the ability to create his/her own general negotiating agents.
 Alrayes, B., Kafalı, Ö., & Stathis, K. (2018). Concurrent bilateral negotiation for open e-markets: the CONAN strategy. Knowledge and Information Systems, 56(2), 463-501.