Human vs. Computer Behaviour in Multi-Issue Negotiation

Contains fulltext : 55027.pdf (publisher's version ) (Closed access)This paper presents two experiments that contribute to the comparison of human- versus computer behaviour in the domain of multi-issue negotiation. The experiments are part of an ongoing endeavour of improving the quality of computer negotiators when negotiating against human negotiators. The validity of the experiments was tested in a case study of closed multi-issue negotiation involving the ABMP negotiation software agents. The results indeed reveal a number of strengths and weaknesses of the ABMP agents. For example, the fairness of deals in negotiations performed purely by ABMP agents is better than the fairness of deals in the comparable negotiations in which humans were involved. Furthermore, in mixed negotiations (i.e., involving human- and software agents) the humans outperform the software agent with respect to the individual performance. Based on the results of the experiments, several suggestions are made to improve the ABMP agent's performance