Research On Protocol And Strategy Of Automated Negotiation

Negotiation is the key tache in commerce activities,and its result directly affects the transaction cost and purchase efficiency.With the development of computer and network technology,commerce action on the Internet becomes more and more pervasive,and traditional negotiation of face-to-face style has already been not able to meet the practical need of applications.Agents can negotiate on the Internet autonomously on behalf of humans and obtain more satisfying negotiation effect.So, the negotiation style of face-to-face are growingly evolved in the negotiation between software agents.In real life,the one-to-many negotiation with multi-attribute happens frequently,and the rapid development of E-Commerce therefore will necessarily call for the automation of the electronic negotiation process,especially the process of the one-to-many negotiation.The main contents and fruits of this thesis are outlined as follows:Firstly,the negotiation framework of one buyer agent and many seller agents is studied.The concept of the one-to-many automated negotiation and related key technology is outlined first,and then a negotiation framework of one buyer agent and many seller agents is defined,furthermore the architecture of buyer agent and seller agent as well as the communication form between them are proposed.Secondly,the problem of the description of the interaction negotiation protocol ontology using OWL is studied.Considering the characteristics of share and reuse and the dynamic of the negotiation process,a Negotiation Protocol Process Ontology (NPPO) is proposed,NPPO is used to describe the contract net interaction protocol with which the one-to-many automated negotiation conformed,and therefore the contract net interaction protocol ontology is gained.Due to the corresponding relationship between SHOIN~+(D) and OWL DL,SHOIN~+(D) is used to describe the contract net interaction protocol ontology and the related OWL class file is gained by using the ontology editor.Thirdly,the one-to-many negotiation strategy with multi-attributes under time constrains is studied.According to the offer law of different negotiation attributes,the negotiation intervals are divided into two kinds.Through the introduced negotiation decision function,the relationship between time constrains and the offer of negotiation attributes is analyzed.As a result,the one-to-one negotiation strategy with multi-attributes under time constrains is established.Based on the analysis,considering the offer game of different negotiation intervals of the sellers in the one-to-many negotiation,the negotiation strategy with multi-attributes under time constrains of buyer and seller is presented respectively.Fourthly,the one-to-many negotiation strategy with multi-attributes considering the alternation learning among agents is studied.On the basis of the established negotiation protocol and the negotiation strategy,through learning form the offer of the opponent,the agent modifies the belief estimate of its acceptable intervals.Then, according to the average expectation of Q at each offer phase of negotiation attributes of the agent and the offer sequences of both negotiators,the multi-attributes negotiation strategy with learning abilities is given.Fifthly,an applicable instance is studied.Based on the above research on the one-to-many automated negotiation protocol and negotiation strategy,under the background of the acquisition of a certain defense key lab,an applicable instance is given,upon which validates the feasibility and validity of theory and methodology of this thesis is validated.