MULTI-AGENT NEGOTIATION MODEL BASED-ON

ARGUMENTATION IN THE CONTEXT OF E-COMMERCE

Guorui Jiang , Yangwei Xu and Ying Liu

School of Economics & Management, Beijing University of Technology,100 Pingle Yuan, Beijing 100124, China

Keywords: Argumentation, Auto-negotiation system, Multi-agent, Negotiation model.

Abstract: In the e-commerce transactions, there are lots of commodities with the same name, but anyone of these

commodities have certain attributes which differ itself from others. During the traditional process of

multi-agent negotiation, only one commodity can be selected as the negotiation object from these

commodities with same name, if buyer agent want to find an appropriate commodity, the flexibility and

efficiency of multi-agent negotiation would be low. This paper studies the multi-agent negotiation model by

argumentation for a group of commodities. It firstly defines all kinds of negotiation elements, then

establishes a negotiation model based-on argumentation and describes the negotiation agreements and

strategies, and finally an example would be presented for testifying the effects of this model.

1 INTRODUCTION

Since the 90s of last century, along with the

development of e-commerce, multi-agent

negotiation is becoming a hot research topic.

Multi-agent negotiation based-on argumentation is

an important kind of automatic negotiation, through

it agents can take part in the process of negotiation

more flexibility and affect other agent’s beliefs,

desires or objectives (Rahwan et al, 2003.

In the field of multi-agent negotiation based-on

argumentation, Jennings and others presented a

model based-on argumentation for multi-issue

negotiation (Jennings et al, 1998); Jing-hua Wu

studied the encouragement model in the multi-agent

negotiation(Wu et al, 2006). However, these papers

are almost concentrated in abstract models. In this

models, the negotiation object is always just one

commodity, and argumentation content is usually not

commodity’s attributes, such as threat and reward. In

the researches of multi-issue negotiation, all the

issues are defined in advance (Wu et al, 2008; Gu et

al, 2010). In the e-commerce transactions, there are

lots of commodities with the same name, but anyone

of these commodities have certain attributes which

differ itself from others. During the traditional

process of multi-agent negotiation, only one

commodity can be selected as the negociation object

Supported by Chinese National Science Fund (71071005,

70940005).

from these commodities with same name, if buyer

agent want to find an appropriate commodity, the

flexibility and efficiency of multi-agent negotiation

would be low.

This paper divides the pricing factors of these

commodities into two classes: main attributes and

secondary attributes. The main attributes is the

common attributes of these commodities and

determine the general price, and the secondary

attributes affect the range of price fluctuation of

these commodities. Secondary attributes would be

considered as argumentation objects, and agents

select certain commodity by argumentation. This

paper firstly defines all kinds of negotiation

elements, then establishes a negotiation model

based-on argumentation and describes the

negotiation agreements and strategies, and finally an

example would be presented for testifying the effects

of this model.

2 MULTI-AGENT NEGOTIATION

MODEL BASED-ON

ARGUMENTATION

2.1 Assumptions

This paper supposes that there are three participants:

seller, buyer and third-party. The third-party plays a

coordinating role and is responsible for the

422

Jiang G., Xu Y. and Liu Y..

MULTI-AGENT NEGOTIATION MODEL BASED-ON ARGUMENTATION IN THE CONTEXT OF E-COMMERCE.

DOI: 10.5220/0003176604220425

In Proceedings of the 3rd International Conference on Agents and Artiﬁcial Intelligence (ICAART-2011), pages 422-425

ISBN: 978-989-8425-41-6

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

establishment of trading relations, the proposal

forwarding and the arbitration.

2.2 The Relevant Definitions

2.2.1 Definition 1

Set of Negotiation Participants:





,,SBA

, S

indicates the Seller Agent, B indicates the Buyers

Agent and A denotes the Third-party Agent.

2.2.2 Definition 2

Negotiation Issue:





II

, negotiation issue

consists of two sub-issues: the main issue and

argumentation issue. The main issue is price. The

argumentation issue

refers the secondary

attribute, and





11 12

, ,,,,,

kkn

II I I   

2.2.3 Definition 3

Price Range and Bid Range: The Price Range is

min max





, it is given by the seller.

min

is the

seller's reservation value and not known to the

outside world; Bid Range is

min max





, it is

given by the buyer.

max

is the buyer's reservation

value and not known to the outside world.

2.2.4 Definition 4

Value of Argumentation Object: Argumentation

object is corresponding to the secondary attribute of

goods. Argumentation object is inherently valuable.

Value of argumentation object is

max min

()







is the weight of secondary attribute of number k,

in addition







2.2.5 Definition 5

Utility: Utility is negotiating expectation or

assessment of opponent’s offer. The utility can be

divided into two classes: expectation utility and

negotiation utility. The expectation utility is

depended on three factors: Price Range, Bid Range

and argumentation. Negotiation utility is related to

times of negotiation. In round

, the negotiation

utility of seller is

min

max min







, and the

buyer’s is

max

max min







For a certain commodity with m argumentation

objects, the seller's expectation utility is:

ma x mi n ma x mi n

ma x mi n

()( )()

SS SS SB

ij ij

PP PP







 







(1)

Among them,



and



are the weights, in

addition









is the number

secondary attribute,







1 mM

the amount of argumentation objects.



is the

number

impact factor of the number

secondary attribute,



1,1 , 0

ij ki











is the amount of values of the number

secondary

attributes.

For a certain commodity with m

argumentation objects, the buyer's expectation

utility is:

(2)

2.3 Model Description

Multi-Agent negotiation model based-on

argumentation is defined as a seven-tuple





,,,,,,,,

OIR TSPB

. In the seven-tuple, A

means negotiation agent, O means the negotiation

object, I is the negotiation issue, R is the range of

Agent’s price offer,



is the space of negotiation

issue, T indicates the times, S is the negotiation

strategy, P is the negotiation agreement, and finally

B is agent’s behavior.

maxmin maxmin

max mi n

()( )()

BB BB S B

ij ij

PP PP















MULTI-AGENT NEGOTIATION MODEL BASED-ON ARGUMENTATION IN THE CONTEXT OF E-COMMERCE

423

3 NEGOTIATION AGREEMENT

Negotiation agreement is the standard of behavior

which agents must comply with when they

communicate with each other. This paper uses this

agreement as the following steps:

(1) The Buyer Agent and the Seller Agent send the

initial prices to the Third-party Agent at the same

time. The Third-part Agent forwards their prices and

determines a certain commodity at random, and then

informs the result to Buyer Agent and Seller Agent.

(2) Buyer Agent and Seller Agent evaluate the

opponent price and confirm that whether accept

opponent proposal or not, and send the their results

to the Third-party Agent; Third-party Agent

judges :if anyone of them accepts opponent proposal,

then process turns to the Sept 3, if there is no

acceptance, then process turns to the Sept 4.

(3) If there is only one acceptance, the Third-part

Agent informs that deal can be done; if there are two

acceptances, the Third-party Agent informs that deal

can be done according to the price offered by last

biding agent.

(4) If there is no acceptance, the Buyer Agent and

Seller Agent evaluate the commodities which have

different secondary attributes according to the

existing prices.

a. If no one has achieved the expectation utility,

both sides continue to offer new prices. Process

turns to the Step 1.

b. If one party achieved its expectation utility, that

UU

, then the agent

will send initiative argumentation for the commodity

which has m secondary attributes. The

argumentation would be evaluated by the opponent

agent, if it be accepted, deal can be done, if not,

process turns to the Step 4. If both sides send

argumentations at the same time, the last

argumentation would be selected by Third-party

Agent.

(5) If deal is done, the negotiation is over, if any

agent refuses negotiation or the negotiation exceeds

maximum time, then the negotiation would be

stopped.

4 NEGOTIATION STRATEGY

Negotiation strategy is what behaviors should be

taken in the process of negotiation. By these

behaviors agent can achieve its max utility. In other

words, negotiation strategy is how to provide price

(Wang et al, 2009).

In this paper, concession strategies are adopted.

min

is minimum price offered by agents, and

max

the maximum price. T is the maximum time

negotiation allows, t is a time variable,



is an

index. Specific concession strategies are defined as

follows:

min max min

() ( )

Pt P P P



 













（





）,The

smaller of attribute value, the better;

max max min

() ( )

Pt P P P





 





（





）,The

bigger of attribute value, the better.

When

0 0.5







negotiation strategy belongs to

impatient type; when

0.5 2





, negotiation

strategy belongs to moderate type; when





negotiation strategy belongs to economical type (Li

et al, 2008).

5 EXAMPLES

There is a category of flash disk. Its brand is M, the

type belongs to N, and Type N has three kind of

color: red, blue and gray. M and N are the main

attributes, color is secondary attributes.

It is assumed that the Pricing Range of S is

[220,312], the impact factor of color



has

three values: 1, 0, -1, corresponding to red, blue and

gray, the weight of color is 0.1. The Bid Range of B

is [226,305], the impact factor of color



also

has three values: 1, 0, -1, corresponding to red, blue

and gray, the weight of color is 0.08. Negotiation

strategy of both sides is to take a moderate type, and

both the indexes



are 0.8. When B start to bid, it

select commodity in random. Now expectation

utility of both sides can be computed, results are

shown in the Table 1.

Table 1: Agent’s expectation utility.

Color

Participant

S B

Red 0.59 0.40

Blue 0.49 0.48

Gray 0.39 0.56

ICAART 2011 - 3rd International Conference on Agents and Artificial Intelligence

424

Figure 1: The bargaining process of negotiation.

The bargaining process is shown in Figure 1.

According to negotiate agreement, in the Round 9,

the pricings of B and S are 256.15, 276.89, deal can’t

be made. However,

0.41

U 

red

UU

Pursuant to the agreement, B would send

argumentation initiatively.

SB (red, 276.89) 

, in

this condition, the negotiation utility of S is 0.59

which is equal to expectation utility, so S would

accept the argumentation of B, deal can be

done.Conclusions can be got from analysis: if

argumentation is abandoned, deal can be done in

round 13. From here we can see that agents send

argumentation initiatively in appropriate time

according to their expectation utility, not only

satisfactory solution would be got as soon as

possible, but also a group of commodities can be

negotiated in one process.

6 CONCLUSIONS

In the field of multi-agent negotiation, very few

researches on negotiation based-on argumentation

are for groups of commodities. This paper researches

the multiple commodities negotiation by

argumentation mechanism, established negotiation

model and testifies it by an example. It proves that

blinding negotiation processes of many commodities

into a uniform multi-agent negotiation process by

argumentation is good selection. As the complexity

of commodity transactions, researches on

multi-agent negotiation based-on argumentation

need to be further excavated.

REFERENCES

Rahwan I, Ramchurn S D, Jennings N R, et al.

Argumentation-based negotiation [J]. The Knowledge

Engineering Review, 2003, 18(4):343-375.

Jennings N R, Parsons S, Noriega P, et a1. On

argumentation-based negotiation [A]. Proceedings of

the International Workshop on Multi-Agent Systems

(IWMAS98) [C]. Boston. MIT. 1998. 1-7.

Wu Jing-hua, Jiang Guo-Rui, Huang Ti-yun. Research on

Models of Reward in Argumentation-based

Negotiation of Agent [J]. Computer Engineering and

Applications, 2006, (36):172-175.

Wu Jing-Hua, Jiang Guo-Rui, Sun Hua-mei et

a1.Modeling and Implementing the Course of

Agent-based Argument-Negotiation [J]. Journal of

Industrial Engineering/Engineering Management. 2008,

(3): 69-74.

Gu Chuan-Long, Sun Hua-Mei, Jiang Guo-rui et a1.

Research on Collaborative Negotiation Model and

Prototype System in Supply Chain [J]. Journal of

Industrial Engineering/Engineering Management. 2010,

(01): 65-69.

Wang Ying, Jiang Guo-Rui, Huang Ti-yun.Two Stage

Negotiation Model for Multi-issue between

Multi-agent [J]. Application Research of Computers.

2009,(2): 573-576.

Li Ran-ran, Sun Hua-mei, Jiang Guo-rui et a1.A Research

on the One-to-many Automated Negotiation Model

Adopting Elimination System Based on Multi-agent [J].

China Journal of Information Systems. 2008, (5):

29-36.

100

150

200

250

300

350

1 3 5 7 9 11131517192123252729

Price

Time

MULTI-AGENT NEGOTIATION MODEL BASED-ON ARGUMENTATION IN THE CONTEXT OF E-COMMERCE

425