Bargain strategies for agent automated negotiation in an e-business environment

Abstract

Abstract. An automated negotiation environment, in which agents employ different bargaining strategies is described. During negotiation, as more information is exchanged in the negotiation rounds, the agents can change the preferences for certain attributes of the negotiation object. The multi-agent system is developed for a real estate agency business model and several use cases scenarios, using intelligent software agents, are implemented.Key words: automated negotiation, multi-agent system, negotiation strategy AMS subject classifications. 68T42, 68T05, 68T351. Introduction. Negotiation is an important issue in business environments. Automated negotiation is a process, in which software agents communicate between them, in order to reach a mutually acceptable agreement [1]. An intelligent agent should be able to negotiate with other agents, which have different negotiation behaviors [2]. Best outcome may be obtained if the agent is able to adjust its strategy, predict or guess the strategy of the other agent [3], or choose an adequate strategy, according to the negotiating partner. Different approaches have been proposed, including machine learning approaches, which can be used to change the agent strategy during negotiation, in order to obtain better results and increased payoffs [4].In our previous works regarding the automated negotiation process, we have proposed bargaining strategies that are based on agent profiles, which can describe statically or can develop dynamically the agent preferences for certain attributes of the negotiation object [5,6]. Using these profiles, agents obtain better results than in the case when fixed negotiation strategies are employed. This paper extends our previous work and proposes an agent model, in which the agents apply the negotiation strategy best suited to them, according to the negotiation situation. A multi-agent environment for automated negotiation is designed, offering services for a real estate business model. In the multi-agent system, there are buyer and seller agents and also a facilitator, which is used when a new agent enters into the system, for registering its services. The agents are designed according to the BDI (Belief-Desire-Intention) model [7]. Each agent has a set of goals, selected from the set of desires.Several use cases scenarios evaluate the negotiation performances. The bargaining takes place in several rounds, before an agreement or a rejection is concluded. In each negotiation round, based on the values and preferences specified by the buyer for the multiple attributes of the negotiation object, the seller agent makes the best possible offer. The agents use linear and non-linear negotiation strategies, which help them in time to increase the gain.The paper is organized as follows: Section 2 presents the negotiation environment, that is our proposed framework for the negotiation system in an open environment. The agent behavior and how each agent acts during the negotiation process is described in Section 3. In Section 4 is applied the develo...

Supply Chain Scheduling Using Double Deep Time-Series Differential Neural Network

“…The data set includes: the number of suppliers m, the number of pending supply chain orders n, and the pending supply chain sub-order processing time data set. Collect historical supply data of the enterprise, and define the supply chain task for the supply demand of product manufacturing: decompose the supply chain task into multiple (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20) supply chain orders, and each supply chain order contains multiple sub-orders (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20), each sub-order is completed by a supplier to complete the construction of the supply chain task data set. In addition, in order to increase the sample size of the data set, the simulation data set is also constructed with reference to the historical supply data of the company to randomly generate the simulation data set, that is, the historical supply data of the company is referred to, and the corresponding processing time curve of the sub-orders of the historical pending supply chain is randomly generated.…”

“…In terms of research status at home and abroad, Pan A et al [8] constructed a "production-scheduled" collaborative process based on multi-agent downstream manufacturers and upstream suppliers; Wang G et al [9] used ontology-based methods to study the supply and demand of resources and services in the supply chain; Serban R et al [10] studied the effect of implementing a differential negotiation strategy on the agent of supply and sales tasks in an e-commerce environment; Jihang Z et al [11] transferred reinforcement learning by reusing the knowledge of the opponent's behavior learned before, and promoted the negotiation between the collaborative tasks of the supply chain; Daniel, J.S.R. et al [12][13][14] use genetic algorithms to optimize supply chain scheduling problems; Ma Yuge et al [15][16] combine particle swarm optimization with supply chain scheduling problems to quickly obtain convergence results.…”

Introduction to the Special Issue

“…Even if different requirements exist for the implementation of the negotiation process in these environments, usually agents-based systems are considered for the implementation of an automated negotiation solution. Even if the results offered by Radu and Florea in [1] are set in relation with an e-business environment, and applied for a couple of usage scenarios, they could be easily expanded for the requirements of clouds.…”