Sven Apel, Gunter Saake
Otto-von-Guericke University, Magdeburg, Germany
Sebastian Herden, Andr
e Zwanziger
Otto-von-Guericke University, Magdeburg, Germany
Agents, Content Addressable Network (CAN), decentralized service-oriented architectures, e-business, Medi-
ator, Partner Relationship Management.
Partner Relationship Management Systems are implemented in many companies to improvethe relations to im-
portant partners along the value-chain. The goal is to differentiate themselves from competitors and their prod-
ucts. That leads to different types of relationships between customers, suppliers and other market-participants.
Some of these types require a mediator, that connects the partners, if a direct connection is impossible. This
article introduces the ideas of Partner Relationship Management within E-Business, and proposes a system
called Mediator, which is able to tie together business partners in a distributed environment. This article
presents the two parts of the mediator (communication system and information system) as well as their key
technologies (an agent-based system based on a Peer-to-Peer network).
In many lines of business, sales departments have
been aimed at products and were assisted by mass-
marketing activities. But the resulting impersonal
contact to customers turned to be ineffective (Ritter,
2003). A reason for that was the customers’ behav-
ior, to change former accepted business connections
quickly. It has been made easy by trends toward glob-
alization and by increasing substitutability of prod-
ucts in mass-markets. This development is going to
be stopped by differentiation in products and in per-
sonalizing the contact to customers.
Literature about Customer Relationship Man-
agement (CRM) and Supplier Relationship Man-
agement (SRM) proposes to utilize the Inter-
net as one communication-channel in Relationship-
Management. Doing so, the complete Customer-
Buying-Cycle can be covered. In the sense of E-
Business it means that the integration of both com-
panies should be so uninterruptive in media-selection
as possible. Hence a mediator is needed, which can
reflect the relationship between both companies and
can pick up and support the benefits of both ideas of
E-Business means, the integrated execution of
all business-processes in a company which can
be automatized, with the help of information-
and communication-technology (I&C-Technologies).
This definition shows that E-Business on the one
hand influences transaction-costs through integration
of business-processes. On the other hand it affects the
pass-through time and respectively the transmission-
costs by the automated execution of business-
processes with information- and communication-
technology. Because of integration first takes place
on the organizational site, which can, if techno-
logically possible, be automated with the help of
I&C-Technologies, a successfully realization of E-
Business is bound on organizational adjustments,
e.g., process-, task- and data-integration (Herden and
Zwanziger, 2004). Therefore E-Business could in
fact generate potentials of success, but they can only
be implemented through restructuring organizational
workflows (Thome, 2002). This is well-founded in
the theory of informational added-values (Kuhlen,
A company has relations along the value-
chain, from pre-suppliers, via direct suppliers, to
distribution-partners, business-customers and end-
Apel S., Saake G., Herden S. and Zwanziger A. (2005).
In Proceedings of the First Inter national Conference on Web Information Systems and Technologies, pages 427-430
DOI: 10.5220/0001229904270430
customers. Additionally companies build up good re-
lations to competitors too, because of changes and
trends in markets, e.g., standards etc. (Riemer and
Klein, 2002).
The choice of the appropriate business-partner and
the exhaustion of potentials of success are realized
by Partner Relationship Management (PRM). By that,
steps in planning, realization and controlling of in-
terorganizational relations to business-partners, with
the aim to save and improve the competitiveness of a
company are understood (Riemer and Klein, 2002).
2.1 Mediator
The mediator presented in this paper, is a technical
system, which has to support the user, e.g. a company
at E-Business. It should depict a market, on which
market participants shall trade goods. Simon explains
a market as an artifact (Simon, 1997; Sunder, 2004),
a tool made by humans in social evolution or through
construction, to trade goods. That market owns an
inner and an outer environment. The rules and struc-
tures of the market lie inside. That includes a consis-
tent language to exchange messages with other par-
ticipants, a mechanism to define and implement the
distribution of these messages, and rules that define
messages, which are valid in each state of the market.
The outer environment is made by the market partic-
ipants, who are defined by their preferences, decision
rules and their endowment.
Therefore, the mediator is made of partici-
pants (suppliers and customers), which maintain
communication-connections, to cover the Customer
Buying Cycle. It is split into a communication system
(inner environment) and a set of information systems,
which are doing information processing for each par-
ticipant (outer environment).
The inner environment of the market is imple-
mented as a Peer-to-Peer network. This architecture
allows to distribute market participants on different
computers over the network (e.g. the Internet). The
outer environment of the market is based on agent-
technology. Agents act on behalf of their users and
shall support users in decision making (e.g. to select
the right partner). Every user can communicate with
her/his agent over different communication channels
(e.g. mobile and stationary devices). Figure 1 shows
a conceptional overview of the mediator system.
To participate in the market, the agent needs knowl-
edge about the inner environment that is realized as an
interface toward communication system. Analogous
to humans’ information processing, agent’s behavior
is also located in meta-cognition and is controlled by
the Application Processor. Rules for decision mak-
ing, preferences and endowment are located within
this meta-cognition and form the outer environment
of the market.
Sensor Effector
Application Processor
Working Memory
System (Agent)
System (Network)
Figure 1: Mediator concept overview
As one can see in Figure 1, the knowledge is lo-
cated in a database or data-warehouse, which is con-
nected to a ERP-System. Therefore, the mediator can
be integrated into current information systems. Doing
so, the areas of analytical, operative and communica-
tive PRM are covered, as well.
In the following sections the communication sys-
tem and the information system are covered.
As mentioned in Section 2.1 the communication sys-
tem is realized as a Peer-to-Peer network. From the
technical point of view a Peer-to-Peer network avoids
a single point of failure. Therefore a Peer-to-Peer net-
work inclines a higher reliability and scalability than
centralized market places. If a centralized market
place falls out, no market transactions are possible.
If all market participants are distributed on different
machines in a network, they can turn out without in-
fluencing other participants.
Moreover, there is an organizational reason to use
a Peer-to-Peer network as communication infrastruc-
ture. A centralized system must be hosted by an or-
ganizational unit. This organization has the absolute
monopoly to decide who could participate on the mar-
ket and for which price. This monopoly may be used
to charge higher fees from successful participants. In
a fully decentralized Peer-to-Peer network all partici-
pants are having equal rights.
Beside the technical and organizational advantages
of a Peer-to-Peer network, it must be possible for par-
ticipants to find other participants who may be a pos-
sible partner. One basic system functionality is, that
participants must be able to offer their products to oth-
ers. In this mediator, each participant is able to offer
the products through a service type definition. A ser-
vice type is a description of a product, which can be
published to other market participants. The service
type defines parameters and their values for the com-
munication between market participants. Due to the
requirement that this mediator cannot contain a cen-
tralized directory, the Peer-to-Peer network must en-
able this functionality.
Content-Addressable Networks (CANs) (Rat-
nasamy et al., 2001) address these issues and provide
a fully decentralized and scalable infrastructure to
distribute a directory. They are able to manage
huge sets of (key,value)-pairs and cope with very
high numbers of parallel transactions. CANs are
using multi-dimensional vector spaces in which data
objects can be stored.
9 8
service type
service provider
Figure 2: A two-dimensional CAN for discovering services
and service providers
Figure 2 depicts a two-dimensional CAN for dis-
covering services and service providers. Service
provider SP
offers services for service type A and
B. Service provider SP
offers only the service type
A. Due to the same service type, the place of these
services A in the directory of the CAN is on a vertical
line. If a search for service type A is started, the rout-
ing through the directory follows this line. All peers
which have an appropriate service type entry in their
directories send back the provider and its address to
the requesting peer.
In Section 2.1 we said, that another part of the com-
munication system is to define rules and a consistent
language for the participants. Figure 3 shows an UML
activity diagram for the flow of messages between
two communication partners (R is the requesting peer,
P is the service provider).
The communication starts with a Request message
of Peer R. The service provider can accept this request
message or might decline the message with a Reset
message. In the case that the provider accepts the re-
questing peer, it responses with a Offer message. If
the offer satisfies the needs of R, it can send a Booking
message to buy the product. Is the offer is not satis-
fying the needs of the requesting peer, it could either
send a Cancel message to cease the communication
or another Request message, which contains a new,
and more detailed, request. If the booking message is
successful, the provider sends back an Ack message.
[R is not
[R cancels
[P does not accept]
[P accepts]
[R is not interested]
[R accepts offer]
[booking ok]
[booking fails]
Figure 3: Message flow in the communication system
When the booking fails the provider sends an Error
This Section covered the network and the directory
of the mediator. We have shown, how market partici-
pants communicate with each other. The next section
describes, how agents are structured and which tasks
the agents must fulfill.
In Section 2.1 we mentioned, that the information sys-
tem is implemented as an agent-based system. Agents
act on the behalf of a user (Maes, 1994; Wooldridge
and Jennings, 1994) and shall support users of the
mediator in decision-making. In the mediator, agents
play two roles: On the one hand, agents shall support
the user in the role as buyer of a product. In this case
they have to implement tasks in SRM. They must be
able to search offers, to rate these offers and to or-
der products. On the other hand, agents appear in the
role of a seller and therefore have to realize tasks from
the CRM. Here, they must be able to submit and to
confirm an order. Moreover, they are able to analyze
other agents, so that the offers are suitable to the re-
questing agent.
Market and
User LogicLogging
User User
User Interface
Experiences with
Business Partners, Reputation
Experiences with Users,
Figure 4: Agent functionality
These requirements lead to the schematic structure
of an agent, which is shown in Figure 4. An agent logs
its experiences with other agents and the user. These
data are analyzed to evaluate the agent’s environment.
The logged data from other agents are used to assess
the trust in other agents. This can be accomplished
by reputation mechanisms (Chavez and Maes, 1996;
Padovan et al., 2001). The log data of the user can be
used to refine the rating mechanisms for offers. This
leads to a personalized analysis of incoming offers on
client-side. All offers from different provider agents
were rated by the requesting agent and displayed as
sorted list. These mechanisms allow to evaluate the
offers with user specific data, which were not part of
the original request. Therefore sensitive data of the
requesting agent will not be published for provider
agents. Both logging variants influence the decisions
made in the market and user logic, which is part of the
Application Processor shown in Figure 1.
This paper revealed the concept of E-Business in the
field of Partner Relation Management (PRM). The
PRM of a company includes the relationship manage-
ment to suppliers (SRM) and to customers (CRM).
The SRM and CRM activities of a company can be
seen as interfaces of a company to a market. To auto-
mate business processes in loosely coupled relations
between organizations a mediator is needed.
The presented mediator is split in an inner environ-
ment (the communication system) and an outer en-
vironment (the information system). The communi-
cation system ties together distributed business part-
ners and provides a decentralized service-oriented ar-
chitecture based on a CAN. The information system
is implemented as agent-based system. Each agent
represents one company on the market and is able to
perform business processes. The use of the media-
tor is domain independent, which means that different
products can be offered and requested.
The concept of this mediator was successfully im-
plemented in a case study in the tourist market.
Different services (e.g. hotel reservation, restaurant
searches and tourist information) were offered by dif-
ferent providers. Users are able to access the offered
information over different devices and to customize
their agent for their needs.
In future work we will investigate in the combina-
tion of different service types. The results of one ser-
vice type shall be the input of another service type.
For instance, a business trip by car includes an ho-
tel reservation and information of petrol stations. The
goal is, that the user only defines her/his destination
and the necessary actions (routing, petrol station plan-
ning and hotel reservation) were done automatically
by the user agent.
Chavez, A. and Maes, P. (1996). Kasbah: An Agent Market-
place for Buying and Selling Goods. In Proceedings
of the First International Conference on the Practi-
cal Appication of Intelligent Agents and Multi-Agent
Herden, S. and Zwanziger, A. (2004). Integrierte Plattform
auf Basis von Agenten- und Peer-to-Peer-Technologie
zur Unterst
utzung des E-Business. diploma thesis,
Otto-von-Guericke University of Magdeburg, School
of Computer Science, Dep. of Techn. and Bus. Inf.
Syst. Business Information Systems I.
Kuhlen, R. (1995). Informationsmarkt: Chancen und
Risiken der Kommerzialisierung von Wissen, vol-
ume 15 of Schriften zur Informationswissenschaft.
atsverlag Konstanz GmbH.
Maes, P. (1994). Agents that Reduce Work and Information
Overload. Communications of the ACM, 37(7):31–40.
Padovan, B., Sackmann, S., Eymann, T., Pippow, I., and
Eggs, H. (2001). Automatisierte Reputationsver-
folgung auf einem agentenbasierten elektronischen
Marktplatz. In Information Age Economy, pages 517–
530. Physica-Verlag.
Ratnasamy, S., Francis, P., Handley, M., Karp, R., and
Shenker, S. (2001). A Scalable Content-Addressable
Network. In Proceedings of the ACM SIGCOMM
Conference, pages 161–172. ACM Press.
Riemer, K. and Klein, S. (2002). Supplier Relation-
ship Management - Supplier Relationships im Rah-
men des Partner Relationship Managements. In Knut
Hildebrand, editor, Supplier Relationship Manage-
ment, volume HMD 228 of Praxis der Wirtschaftsin-
formatik, pages 5 – 22. dpunkt.verlag.
Ritter, U. (2003). Multi-Channel-Management als Dif-
ferenziator am Markt. In Helmke, S., Uebel,
M. F., and Dangelmaier, W., editors, Effektives
Customer Relationship Management; Instrumente -
uhrungskonzepte - Organisation, page pp. 195.
Gabler Verlag, 3 edition.
Simon, H. A. (1997). Empirically Grounded Economic
Reason, volume 3 of Models of Bounded Rationality.
The MIT Press.
Sunder, S. (2004). Markets as Artifacts: Aggregate Effi-
ciency from Zero-Intelligence Traders. In Mie Augier
and James March, editor, Models of a Man: essays in
memory of Herbert A. Simon. The MIT Press.
Thome, R. (2002). E-Business. Informatik Spektrum, 25
Heft 2:pp. 151.
Wooldridge, M. and Jennings, N. R. (1994). Agent The-
ories, Architectures, and Languages: A Survey. In
Wooldridge, M. J. and Jennings, N. R., editors, Proc.
ECAI-Workshop on Agent Theories, Architectures and
Languages, pages 1–32, Amsterdam, The Nether-
lands. Springer, Berlin u. a.