Automated Negotiation Strategies Prediction in Multi-agent Systems

The paper presents different negotiation strategies needed for reaching a mutually acceptable agreement between providers and consumers that are modeled as software agents. The focus is on designing, implementing and evaluating different complex strategies, based on the negotiation context and multiple communication protocols between the agents. For the strategies to become successful, it is essential for the agent to possess information about the other parties and the characteristics of the market. Learning is a defining characteristic for an intelligent behavior, so, with the help of machine learning algorithms, an agent builds an advanced model, using the gathered information from previous actions.