Complex negotiations in multi-agent systems

摘要

In this thesis, we propose automated negotiation models for complex scenarios. More specifically, we focus on (ⅰ) negotiations where agents have limited computational resources, and (ⅱ) negotiations where negotiation parties are formed by more than a single individual (i.e., negotiation teams). In the former case, we propose a bilateral model that obtains statistically equivalent results to models that explore the whole non-linear utility function while using less computational resources. In the latter case, we propose automated negotiation models for a negotiation team that negotiates with an opponent by means of the bilateral alternating protocol. We put a special emphasis on models that are able to guarantee unanimity with regards to team decisions in the negotiation.