Abstract
This paper presents a model of automatic negotiation agents in an open environment. Agents are motivated by the gain they may obtain while fulfilling their goals, but their behaviour can change during negotiation according to previous interactions with other agents in the system. Changing behaviour may refer to either the use of different negotiation strategies or to concessions made for other agents, with which they have successfully negotiated in the past. To this aim, an agent develops a set of partners’ profiles during negotiation: the preference profile, the cooperation profile, and the group negotiation profile. The first two profiles characterize individuals, while in a group negotiation profile, several agent profiles are clustered according to commonly discovered features. Different approaches to the development of these profiles are presented.
Highlights
Negotiation between agents appears in different area of research, like electronic commerce, distributed resource allocation or virtual enterprises
For each tuple (s, a), where s is the internal state of the agent and a represents the rule applied during negotiation, our preference coefficients are updated using a formula, in a similar manner to the Q-learning algorithm: PC s, a PC s, a where max PC s, a is the expected preference coefa ficient of the internal state of the agent s, when applying the rule a . is the learning rate representing the impact of the update value and r(s) is the immediate reward for the internal state of the agent s
We have developed an automated negotiation environment, which combines the agents’ beliefs about the other agents in the system with the possibility to represent and modify the negotiation strategy
Summary
Negotiation between agents appears in different area of research, like electronic commerce, distributed resource allocation or virtual enterprises. Agents are acting in an environment in which other agents may enter or leave, some of them known before, some others encountered for the first time In this context, the design of intelligent agents with a complete pre-defined negotiating behaviour represents a challenge for the designer, especially when the agents are conceived to be general purpose and not limited to a specified domain. The work reported in this article extends the results presented in [3] and describes in detail how different profiles can be built, including the selection of group profile and the clustering of negotiation instances, based on machine learning algorithms. The negotiation profiles the agent uses are: the preference profile, which implements the agent negotiation strategy, the cooperation profile, which improves the agent interaction with the partners, and the group-ofpartners’ negotiation profile, which clusters the profiles of several agents.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
