Design Decision Mechanism Of Automated Negotiation In Electronic Marketplace

Supporting various users to perform effective business activities on Internet is one of main research challenges in electronic commerce area. The development of next generation electronic commerce is required to improve the efficiency, effectiveness and degree of automation of dynamic negotiated trades. Owing to their prominent characteristics of agent technologies, they have been gradually used to support some negotiation stages of online businesses. And agent-based automated negotiation has become one of the newest international research topics. However, until now although many methods have been proposed to model negotiations between autonomous agents, they neglect the requirements of realistic online business environment imposed on automated negotiation as well as the characteristics of agents. Problems how agents make decisions and select reasonable negotiation actions haven't been solved. Aiming at the negotiation decision problems of agents, in this paper the decision mechanism of agents is designed and takes into account all key phases of negotiation decisions.First, according to characteristics of electronic marketplace, the requirements and application conditions of automated negotiation are first analyzed. Then the background conditions of designing automated negotiation are proposed. After analyzing two modes of agent-based negotiation and comparing three negotiation protocols, argument based negotiation protocol is used as the basic negotiation mechanism that agents participate in online negotiations. Negotiation speech acts of agents in such protocol are defined by using dialogue game method. Then the negotiation environment of agents is proposed and based on the environment agent's negotiation decision process is modeled as its belief revision process. Secondly, in order to perform negotiation tasks agents must"understand"their users'real requirements. So user negotiation profile model is proposed to elicit and represent information that relates concrete negotiation tasks and provides necessary information for negotiation decisions of agents. Two-stage incremental elicitation method of user negotiation profile based on cases is proposed and semiring-based soft constraint satisfaction problems are used to represent various user negotiation profiles. Preference constraints and trade-off preference constraints are defined and three basic operations of constraints are provided to realize the flexible elicitation and representation of user negotiation profiles.Thirdly, owing to the autonomous characteristic agent should determine their negotiation goals of different stages. Then agent's BMG (belief-motivation-goal) architecture is proposed and used to specify agent's decision process. Goal is used as the key component of agents to drive their decisions. After defining different types of goals, goal topology is analyzed and goal dynamic is proposed from the three interleaved processes, that is, goal generation, goal selection and goal adjustment. And then in order to determine agent's action focus in different states, meta-level control mechanism of goals based on motivations is proposed. Fourthly, after specifying various goals agents should try their best to achieve them. So negotiation strategy and action selection method of agents are proposed. Because argument-based negotiation protocol is used as the protocol between agents, negotiation actions of agents are composed of negotiation offers and arguments. Local searching algorithm using incomplete preference information of negotiation opponents is proposed to determine concrete negotiation offers. And methods of argument generation, argument selection and evaluation based on argumentation theory as well as the preference belief revision method of agents are provided.In order to improve the acceptance of negotiation outcomes, agents should explain their decision processes. So explanation mechanism of agents and the explanation model based on Toulmin model are proposed. Finally, an example is given to illustrate the whole design.In this paper decision mechanism of negotiating agents is designed following the research line, that is, the delegation of negotiation tasks, task execution, and explanation of negotiation outcomes. Many new ideas and methods are provided. They enrich current researches of automated negotiation and can be applied in developing practical automated negotiation system.