Multi Agent based Hybrid E-negotiation System in E-commerce

Kourosh Dadashtabar Ahmadi, Nasrollah Moghadam Charkari

International Journal of Information Processing and Management. Volume 2, Number 2, April 2011 

Multi Agent based Hybrid E-negotiation System in E-commerce

1
Kourosh Dadashtabar Ahmadi, 2Nasrollah Moghadam Charkari
*1
Tarbiat Modares University, Dadashtabar@yahoo.com
2
Tarbiat Modares University, Charkari@modares.ac.ir
doi: 10.4156/ijipm.vol2. issue2.10

Abstract
E-business negotiation is a decision-making process that seeks to find an electronic agreement,
which will satisfy the requirements of two or more parties in presence of limited information, and
conflicting preferences. Therefore, we need to use hybrid systems that are able to have an appropriate
performance in real conditions of different types of automated negotiations. This paper presents a
multi agent based hybrid system that is using multi criteria decision-making and electronic negotiation.

Keywords: E-negotiation, Decision Making, Multi-agent

1. Introduction
Electronic commerce has changed the view of traditional business trading behavior. To automate
most of commerce time-consuming stages of the buying process, software agent's technologies have
been proposed and employed in different transaction stages of e-commerce. The agents in e-negotiation
dialogs based on their owner requirements until reaching agreement on one or multi issues of the
negotiation [1].In addition, in many real conditions negotiation, negotiation agents have only limited
information about their opponents and bounded rationality. Therefore, using heuristic algorithm to
develop negotiation agents’ decision-making mechanism is useful. For this propose, we proposed multi
agent based interactive recommendation and automated negotiation system (IRANS), so we expect that
the negotiation time and transaction cost are reduced [4]. Also by the system allow negotiators to
negotiate simultaneously in multi negotiation [8]. Since negotiations are often done in the intangible
and complex negotiation space [15], negotiation deadline is limited and information is imperfect [26],
therefore real negotiation models related to this space should be designed and implemented in
corresponding to negotiation goals and constraints. Also, this design should be able to (a) be integrated
with other technologies like data mining, information researching and information retrieval [12] (b)
carry out of goal driven multi negotiations with multiple issues at the same time [11] (c) create user
profiles [10] (d) support vital negotiation mechanism [17] (e) create a hybrid negotiation system
(human, system and agent) [12]. Our proposed system is a new design proportionate to above
discussion while simple price comparison discarded. Moreover with this design while there is not
perfect information of proposal space, negotiation space and goal space, agents can negotiate to each
other according to their appropriate requirements and constraints(goals). This paper is structured as
follows. In next section, we present related works. Section 3 explains new process model according to
the design and roles of multi agents in the design also provides multi agent based architecture of buyer
and seller. In section 4, we explain architecture of multi agents IRANS. Section 5 illustrates multi
criteria decision-making to find best opponent negotiators. Final section gives conclusions and
highlights open issues that need to be addressed in future.

2. Related works
Various e-market places like KASBAH, eBay, Michigan and each electronic markets employed e-
negotiation based on agent technology. Unfortunately, the market agents trade only by price [2].While
in real world, negotiations are not only by price but they often involve multiple issues (e.g., Price,
Quantity and Quality Product) [14]. In Ref [27] a fuzzy similarity based trade-off mechanism is
proposed to search for near optimal negotiation solutions based on realistic assumptions. The solution
based on a bilateral negotiation, whereas negotiation in e-commerce often involves multiple parties.
Our proposed model solved this problem. Ref [25] and [26] has developed a sequential negotiation

- 88 -
model that allows agents to revise their beliefs about the negotiation environment and the preference of
opponents. The multi agent learning mechanism is based on probabilistic framework and employed
Bayesian representation and learning. It was assumed that domain knowledge an offering price given
the opponents reservation price was available to the system. The proposed model is limited to learning
single negotiation issue (e.g., price), also this idea conflict to keep preference information private. Our
proposed model does not assume the availability of the opponent preference information. In Ref [5],
notwithstanding is used analytic hierarchical process (AHP) for C2B multi issues negotiation but
interdependency between criteria is ignored. Also the transaction cost is high. In Ref [13] same to Ref
[5] is used fuzzy AHP decision-making method for B2C negotiation. Therefore these solutions are not
really business conditions. The hybrid negotiation system in this paper is able to solve this subject.
Moreover, our automated negotiation is based on the multi agent system approach where can make
autonomous negotiation decisions.

3. IRANS process model and role of multi agent in proposed model

For design of an e-negotiation system, one important activity is determining negotiation process
model because the process model describes phases and assigns different activities to them [2]. Another
important factor in design of the system is protocol [1]. The protocol is a structure model, involves set
of rules, which dictate negotiation system processing and communication tasks, and constrains on
activities [2]. In addition, process model and protocol provide plausible method for negotiation systems
and remove additional activities of the system [1]. The model with activities related to every phase
presented in “Fig.1”. Negotiation process model is formed by goal driven negotiation and requirements
assigned to agents. Constraints and requirements are under title of negotiation issues then break down
to main goals and due to clear relation between main goals, each main goal, and then break down to
sub goals based on the relations exist between sub goals such as support, interference, and
independence. Then implantable goals are selected from sub goals based on some rules. According to
topology structures of goals, we can define goal driven negotiation process model. Based on such
topology, agents are able to improve their behaviors over time. In that way, they may get better with
experiences at selecting and achieving goals by taking correct actions. So agents can take proactive and
reactive negotiation actions.

- 89 -
 Multi Agent based Hybrid E-negotiation System in E-commerce
Kourosh Dadashtabar Ahmadi, Nasrollah Moghadam Charkari
International Journal of Information Processing and Management. Volume 2, Number 2, April 2011 

Figure 1.Negotiation process model and assigns different activities to them

Now a day's, because in e-market there are infinite participants[18], exponentially growing amount
of information available[13], information transparency, information overloading, different negotiation
mechanism[12] price wars between buyers and between sellers[10]and complexity of modern business
trade [16], therefore multi agent based automated negotiation systems are hopeful technologies that
play an important role[5]. Multi agents systems can adopt various mechanisms (e.g. bidding, auction,
bargaining, and arbitration), information discovery and collection, negotiator selection, proposal
generation and evaluation, play of advisor and provide effective suggestion, fully understand of owner
requirements and preferences, coordinate interdependent relationships between various negotiations,
build comprehensive user models and finally the systems can make trust between both users [9]. In
addition, multi agents allow negotiator to play both roles of buyer and seller at the same time [17]. Our
proposed design has considered the notations and results of this research show that proposed multi

- 90 -
agent architecture would reduce negotiation time and provide rapid response for their owners. Next
step, we explain the particular design of multi agent's architecture. In order to explain the architecture,
first we have to illustrate types of agents employed in the architecture and then we show buyer and
seller agent architecture.
In IRANS purposed model, Owner profile agent is to present owner goals, and aid the negotiator
deciding on goals and strategies [5]. In addition, agents of this type would be able to adapt to the
changes based on owner behavior in the process of negotiations [2, 9]. Searcher Agent is an agent that
searches potential buyers or sellers that are in other distributed environments and performs the role of
managing, querying, or collating information from many distributed sources [13]. A simple rule
matching mechanism is developed to extract relevant negotiator information from the search results.

Figure 2. Buyer and seller agent architecture based on multi agent

This agent also performs post processing for the retrieved items and gives a list of qualified buyers or
sellers with the offer. Information mediator agent would envelope in actively searching, fetching,
filtering, and delivering information relevant to the issues from market knowledge base. Also this agent
type is able to identify the objectives, preferences and strategies of the opponent [2, 10]. Recommender
agent is able to generate a set of likely offers to be considered for submission to the opponent [10]. The
proposed of advisor agent is to evaluate the offers received from the opponent and provide a feedback
on the defect and, possibly benefits of these offers [5, 13]. Negotiator agent is responsible for
negotiating with potential buyers or sellers, with respect to the preferences that have been collected
from a group of participants [10]. Here, the goal is to bargaining with candidate sellers or buyers for
the best offer that satisfies the most demands and preferences of group members. In addition, this agent
may be capable of conducting negotiations by itself in a semi-autonomous or fully autonomous mode
[13]. Applicability of full automation depends on the degree of certainty in objectives, preferences, and
tactics of the negotiator therefore the agent is an agent that optimizes the buyer or seller utility based
upon the requirements from owners’ requirements and constraints [2].Buyer or seller mediator agent is
an agent that delivers the status messages of active services between negotiators agents, and between
peer agents. In addition, this agent can play a role of an expert agent and in a sense; the agent is an
intelligent administrator agent [10].

4. IRANS architecture

For illustrate of the propose design, primarily we describe some assumption and then base on
negotiation protocol (e.g. one-to-one, one-to-many and many-to-many) in this section we presents
multi issue multi agent-based architecture. First assumption is that, the agents have incomplete and
limited information and knowledge of requirements (e.g. preferences, issues and constraints) and

- 91 -
 Multi Agent based Hybrid E-negotiation System in E-commerce
Kourosh Dadashtabar Ahmadi, Nasrollah Moghadam Charkari
International Journal of Information Processing and Management. Volume 2, Number 2, April 2011 

negotiation strategies of the other agents that is means all the information about the negotiation objects
and domain is assumed unavailable. They make decisions in relation to available information about
private issues and individual negotiation strategies. The agents exchange information in the form of
offers [22,25]. An offer is a complete solution that is currently preferred by an agent given its
preferences, attributes and the negotiation history of offers and counter offers. An agreement takes
place when another party accepts a particular offer. During the negotiation process, the offers change
by modifying the value of each issue according to negotiation space that changes dynamically. Each
negotiation agent has an iterative process of evaluating the offers, updating the available believes and
making the counter offers, according to the individual negotiation [4,7]. These heuristic approaches
make use of offer based information which is exchanged between agents to infer useful information
which lead to better agreements [26]. Also in this paper, second assumption is that we follow the
intuition that is agents may exchange explicit meta-information to improve the way they negotiate.
Argumentative messages relevant to the agents' positions are exchanged that can reveal any unknown,
non-shared, incomplete, imprecise information about their underlying mental attitudes [1]. Goal driven
negotiation is a particular type of argument-based negotiation, which can also be seen as a special case
of heuristic approaches. In addition, this approach assumes that parties are much more likely to come to
a mutually satisfactory outcome when the object of negotiation is not considered as central as the
agents' underlying goals [24].

4.1. Bilateral negotiation in IRANS

There are two key points in the field of E-Negotiation systems, protocol, and strategy. The protocol
determines the flow of messages between the negotiating agents; it is necessarily public and open. The
strategy, on the other hand, is the way in which an agent acts within the protocol in an effort to get the
best outcome of the negotiation [3]. In this section, we describe the process of negotiation between the
seller agent, the buyer agent and mediator agents. This type of negotiation is restricted to two
negotiation partners' one buyer and one seller. The mediator agents will behave differently in our
negotiation protocol depending on the types of negotiation. At the beginning, the seller agent evaluates
the profit and chooses an optimal offer to the buyer agent. After receiving the offer from the seller
agent, the buyer agent updates recent belief according to opponent offer, and then estimates the offer to
judge whether accepts it or not [10]. If the buyer agent does not accept the offer, then the buyer agent
will adjust its own strategy, proposes a new offer, and sends the offer to the seller agent. For evaluation
of the offer and counter offer, we can use multi-issue utility theory and fuzzy decision process [20].
The belief is having a preview about the opponent private information and recent environment state.
The agent belief model updates the negotiation according to received offer. Through offering and
counter-offering, the agent would gradually know opponent offer strategy, and then it can propose a
favorable offer by itself. The owner agent will check the restrictions according to the corresponding
issues and restrictions in the database of user’s restrictions, after the negotiation agent received the
offer. If the restriction is disobeyed, the agent should broaden restriction or adjust restriction artificially,
adjust negotiation strategy according to negotiation environment and the received offer, and then
propose a new offer. In the bilateral IRANS as you see in “Fig .3”, there are three important views: (a)
the mediator agent is expert agent and manages communication between IRANS entities (e.g. buyers,
sellers, buyer agents, seller agents). Also the agent performs negotiation process, and filters the offers
from buyers/seller agents on behalf of owners. (b) The seller and buyer agents, which are negotiating
with a large number of agents. They can refuse to negotiate by setting a timeout. This timeout allows
the agents to avoid interacting with overloaded agents. (c) These buyer /seller agents involve multi
agents like “Figure.2.”

- 92 -
 Figure 3. Bilateral Negotiation in IRANS

4.2. Multilateral negotiation in IRANS

The conventional approaches often used to model and analyze negotiation, like auction theory, are
hardly useful for modeling of multilateral negotiation about many issues and with vague preferences
and uncertain information on either negotiating side. There are two types of multilateral negotiation,
One-sided multilateral negotiations that are characterized by one seller and many buyers or by many
sellers and one buyer [3,16]. This type of negotiation is deemed the standard form of auctions. There is
a significant limitation of auction-based negotiation. They do not allow for exchanging offers and
counter offers, and exploiting the flow of information in both directions [10]. Another multilateral
negotiation is double-side multilateral negotiation. Many buyers and many sellers [3] characterize
double-sided multilateral negotiations. This section presents multilateral negotiation that: (1)
Buyer/Seller agent can use different negotiation strategies to negotiate with different agents; (2) Buyer
agent negotiate with seller agents through mediator agents by creating a number of one-to-one
negotiations between each of seller agents. Also Seller agents negotiate with buyer agents through
mediator agents by creating a number of one-to-one negotiations between each of buyer agents “Fig .2”.
Fig .4 shows how a one-to-many negotiation could be implemented by coordinating a number of
parallel one-to-one negotiations. In the complex business world, many negotiators participate in the
negotiation. Each of users is responsible for representing only his/her own goals and interests in the
negotiation. Therefore, to prevent price war between users, we needs to coalition agents. In “Fig. 5”
coalition process illustrated clearly.

- 93 -
 Multi Agent based Hybrid E-negotiation System in E-commerce
Kourosh Dadashtabar Ahmadi, Nasrollah Moghadam Charkari
International Journal of Information Processing and Management. Volume 2, Number 2, April 2011 

Figure 4. Multilateral parallel negotiation

Figure 5. Negotiation between coalition agent and others agents

4.3. Integrated architecture of IRANS

According to the above discussions, in the “Figure6” we present integrated IRANS architecture.
Viewpoint of the architecture is that agents interacts and communicates each other. By this architecture
while there is not complete information and there are multi issue, e-marketplace users negotiate with
each other in real business world.

- 94 -
 Figure 4. Multilateral parallel negotiation

5. Multi criteria decision-making in IRANS

The seller or buyer selection problem is a strategic issue in e-marketplace. On the other hand,
selecting the optimal seller for buyer and buyer for seller among many alternatives is a decision-
making [2]. This paper develops an evaluation model based on Analytical Network Process (ANP) and
the Technique for Order Performance by Similarity to Ideal Solution (TOPSIS), to help the sellers or
buyers in e-marketplaces for the selection of optimal alternatives in fuzzy environments. The fuzzy
environment deals with the uncertainty and imprecision related with information concerning several
parameters [21]. The vagueness is handled with linguistic values parameterized by triangular fuzzy
numbers. The fuzzy ANP (FANP) is used to capture the fuzziness in evaluation of criteria and sub
criteria [23]. FANP has some additional advantages according to the conventional ANP method. After
we obtain criteria weight by FANP, to avoid a large number of pairwise comparisons in ANP method
in this proposed model, we use fuzzy TOPSIS method to obtain the final ranking results.[19, 21].
In a negotiation environment, if there are (m) criteria and (n) alternatives, we need to (k) pairwise
comparisons to run a full FANP solution [19].

 m  n n  1 
k   (1)
 2 

6. Conclusion
In this paper we present a new design that can work in real business world. The proposed
architecture increase probability of reaching to agreement while there isn’t any opponent information
and negotiators face to almost imperfect information. Also through using architecture, a negotiator can
negotiate with several participants, different strategies and multi issues at the same time. To improve
the design, in future we are going to implement proposed design and log result of the implementation.

7. References
[1] H. Al-Sakran and I. Serguievskaia, “A framework for developing based e-negotiation system”,
Journal of Computer Science, Vol 2. No.2, pp 180-184, 2006.
[2] P. Braun, J. Brzostowski1, G. Kersten, J. Baek Kim, R. Kowalczyk1, S. Strecker, and R. Vahidov,
“Intelligent decision-making support systems: foundations, applications and challenges” chapter
15, pp. 271-300, Springer, London, 2006.
[3] R. Buttner, “A classification structure for automated negotiations”, In Proceedings of the 4th
workshop onmulti agent systems in e-business, pp.523-530, IEEE Computer Society, 2006.
[4] Y.M. Chen and P.N.Huang, “Agent-based multi-issue negotiation scheme for e-market
transactions”, Journal Applied Soft Computing, Vol.9, pp 1057-1067, 2009.

- 95 -
 Multi Agent based Hybrid E-negotiation System in E-commerce
Kourosh Dadashtabar Ahmadi, Nasrollah Moghadam Charkari
International Journal of Information Processing and Management. Volume 2, Number 2, April 2011 

[5] D. N. Chen et al, “An agent-based model for consumer-to-business electronic commerce”, expert
systems with applications 34, Elsevier Ltd, pp.469–481, 2008.
[6] C. Cheng et al, “Buyer- supplier negotiation by fuzzy logic based agents”, Proceedings of the Third
International Conference on Information Technology and Applications, 0-7695-2316-1/05 IEEE,
2005.
[7] P.Faratin, C.Sierra, and N. R. Jennings, “Negotiation decision functions for autonomous agents”,
International Journal of Robotics and Autonomous Systems, Vol.24 (3-4), pp 159-182, 1999.
[8] V.Jian and S.G.Deshmukh, “Dynamic supply chain modeling using a new fuzzy hybrid
negotiation mechanism”, In Int. J .Production Economics, No.122, pp 319-328, 2009.
[9] G.E. Kersten, and G. LoAspire, “ Integration of negotiation support system and software agents
for e-business negotiation”, International Journal of Internet and Enterprise
Managment,Vol.1,No.3,pp 293-315, 2003.
[10] T.Kijthaweesinpoon and P.L.Zhou (a), “An automated mechanism for negotiations in e-commerce
systems”, International Seminar on Business and Information Management, IEEE, pp.194-198,
2008.
[11] T.Kijthaweesinpoon and P.L. Zhou (b), “An extended unified negotiation model for e-commerce
systems”, International Symposium on Information Science and Engineering, IEEE, pp.758-761,
2008.
[12] S.Lei and S.Zhang, “Effective automated negotiation in electronic marketplace”, IEEE. 2008.
[13] W.Y. Liang et al, “The agent-based negotiation process for B2C e-commerce”, International
Journal of Expert Systems with Applications, Vol .37, pp 348-359, 2010.
[14] L. Li et al, “Multilateral multi-issue automated negotiation model based on GA”, International
Conference on Intelligent Computation Technology and Automation, pp.85-89, 2008.
[15] S.C. Ng et al, “Intelligent negotiation agents in electronic commerce applications”, In Journal of
Artificial Intelligentce,Vol.2,Issue 1,pp 29-39, 2009.
[16] S.Park and S.B.Yang, “An efficient multilateral negotiation System for Pervasive Computing
Environments”, Engineering Applications of Artificial Intelligence 21, IEEE, pp.633-643, 2008.
[17] Y.K. Raymond, “Towards a web services and intelligent agents-based negotiation system for B2B
e-commerce”, Electronic Commerce Research and Applications 6, IEEE, pp.260–273, 2007.
[18] V.Rudolf, “Learning about Preference in Electronic Negotiation”, In European Journal of
Operational Research, No.194, pp 452-463, 2009.
[19] M.Dagderviren, S.Yavuz and N.kilinc, “Weapon selection using the AHP and TOPSIS methods
under fuzzy environments” An International Journals Expert Systems with Application, No.36,
pp8143-8151, 2009.
[20] S.Boaran and Kerim Goztepe, “Development of a fuzzy decision support system for commodity
acquisition using fuzzy analytic network process” An International Journals Expert Systems with
Application,No.37,1939-1945,2010.
[21] M.Dagderviren and I.Yukse, “Weapon A fuzzy analytic network process model for measurement
of the sectoral competition level” An International Journals Expert Systems with Application,
No.37, pp.1005-1014, 2010.
[22] M.Louta,,I.Roussaki and L.Pechlivanos “An intelligent agent negotiation strategy in the electronic
marketplace environment” European Journal of Operational Research, No.187, pp 1327-1345,
2008.
[23] Z.Ayag, and R.G.Ozdemir “A hybrid approach to concept selection through fuzzy analytic
network process” Computers & Industrial Engineering, No.56, pp 368-379, 2009.
[24] C. Bartolini, C. Preist, and N. R. Jennings, “A software framework for automated negotiation”,
Springer-Verlag Berlin Heidelberg, pp.213.235, 2005.
[25] T. Skylogiannis, et al, “A System for Automated Agent Negotiation with Defeasible Logic-Based
Strategies”, Proceedings of the IEEE International Conferences on e-Technology, e-Commerce
and e-Service, Hong Kong, pp .44–49, 2005.
[26] P.Pasquier et al, “An empirical study of interest-based negotiation” Proceeding of ninth
international conference on electronic commerce” Published online Springer, 2010.

- 96 -