TRANSCENDING TAXONOMIES WITH GENERIC AND

AGENT-BASED E-HUB ARCHITECTURES

George Kontolemakis, Marisa Masvoula, Panagiotis Kanellis, Drakoulis Martakos

Department of Informatics & Telecommunications

National and Kapodistrian University of Athens

University Campus, Athens 15771

GREECE

Abstract: If effectively utilized, modern technologies such as ontologies and software agents hold the potential to

inform the design of the next generation of E-Hubs. In terms of their evolution, we argue that taxonomies

as tools hold the danger of stifling innovation as they may implicitly impose boundaries on the problem

domain. We proceed to use one that is well-referenced in the literature and identify a number of issues that

can be seen as limiting factors, proposing a generic and agent-mediated architecture that holds the potential

of addressing them.

1 INTRODUCTION

E-Hubs bring together buyers and sellers in real-time

trading communities at relatively low cost (Rosson,

2000). Taxonomies are classification systems that

allow one to uniquely identify something. One of the

best known examples is the science of systematics

which classifies animals and plants into groups

showing the relationship between each (Bishop et

al., 1995). Any classification should be presented in

such a form that stakeholders can use to specify the

identity of the system they seek as a solution to a

problem and to prescribe the base functionality that

will guide the design. But taxonomies can also stifle

innovation by limiting the views of the stakeholder,

imposing boundaries through categorization

schemes and levels of abstraction that can and must

be challenged. In this work we take as a starting

point a well-referenced taxonomy for E-Hubs

(Kaplan et al., 2000) and identify a number of issues

that demand our attention.

The next section contains the issues that drive our

researc

h towards the proposition of a generic E-Hub

architecture which is presented in the third section.

The paper concludes by presenting the next steps in

our research endeavors.

2 E-HUB TAXONOMIES

Regarding the taxonomy in (Kaplan and Sawhney,

2000), we have put forward four issues that can be

used for arguing against the underlying assumptions

behind it and in effect question any claims to future

applicability that it may have. Issue 1:

Mougayar

(2000), states that E-Hubs are not as open as they

could or should be. Every company has different

types of products and services and a different

customer base to deal with. Even if they have the

same product categories, they may present them

differently to their clients emphasizing on special

attributes that they only amongst the other hub

participants choose to provide. In addition, the

particular taxonomy differentiates E-Hubs to those

that deal with manufacturing and to those that deal

with operating products. Simply stated, if a

company transcends these categories in the physical

world, it cannot do it in the virtual. Issue 2:

Virtually every hub or marketplace created focuses

on either B2B or B2C business transactions. An

integration of both categories would yield a generic

e-hub made for all stakeholders across the process

flows and covering every step of the way from

production to consuming. It is important to note that

the taxonomy proposed by Kaplan and Sawhney

only covers B2B E-Hubs. Issue 3a:

Categorizing E-

Hubs as in the particular taxonomy may strip away

any flexibility that a prospective participant could

have used in order to evaluate his options to engage

or not. In systematic sourcing the conditions are not

favorable for small participants since they cannot

achieve the same terms and discounts as large users

who buy large quantities through the E-Hub. In spot

sourcing the conditions are not favorable for large

clients since even if they buy a lot from the E-Hub

they sometimes during the auction may end up

buying to a steeper price than a small business.

Accordingly, an E-Hub offering both spot and

systematic sourcing may help to avoid the

297

Kontolemakis G., Masvoula M., Kanellis P. and Martakos D. (2005).

TRANSCENDING TAXONOMIES WITH GENERIC AND AGENT-BASED E-HUB ARCHITECTURES.

In Proceedings of the Seventh International Conference on Enterprise Information Systems, pages 297-300

DOI: 10.5220/0002539202970300

 SciTePress

appearance of phenomena that relate to the ‘chicken-

and-egg’ problem. Issue 3b:

Long-term contracts in

current E-Hub systems are negotiated on fixed

product prices. The negotiation schemes of E-Hub

architectures offer no allowances regarding the

users’ wish and freedom in generating a new offer or

have any influence over the “game rules”. Again, an

integration of both categories can help the user to be

flexible to market dynamics and either harden or

soften his/her negotiation stance to his advantage.

Issue 4:

According to (Kaplan et al., 2000), E-Hubs

can be either neutral or biased. In the physical world

there exists no such distinction. A business can be

neutral to some and biased to others. Buyers as well

as suppliers are needed for the system to function

and neutrality can be decided according to the

chosen tactics and strategies.

In the next section we propose a prototype of a

generic and agent-mediated E-Hub architecture that

was designed so as to confront the above issues

perceived as impediments to the evolution of open

and truly flexible E-Hubs. What we imply is that

our research path followed probes us to question the

applicability of existing taxonomies and revisit their

underlying assumptions.

3 A GENERIC AND AGENT-

ENABLED E-HUB

ARCHITECTURE

To be effective in achieving their set objectives, E-

Hub users must analyze a wealth of information,

negotiate over multiple contracts, and execute a lot

of complex transactions on the Internet

(Kontolemakis et al., 2004). To this end, agents play

a significant role and many systems such as those

proposed by Debenham (2000) and Shen et al.

(2002) are beginning to incorporate them in their

architecture.

Agents are clearly identifiable solving entities with

well-defined boundaries and interfaces and have

evolved from Multi-Agent Systems (MAS).

The main areas related to E-Hubs where agent-based

functionality can be applied are ontologies, advising

services and negotiation. By ontology we mean the

specification of the knowledge structures used to

define concepts and the relationship among those.

The primary focus when designing the ontology

model of an E-Hub is to satisfy those design

requirements that will enable its extension, share and

reusability both within and outside the boundaries of

the hub infrastructure (Albers et al., 1999). The

advising service an agent is to deliver requires the

consideration of a multiplicity of design issues and

parameters such as intent (the goal of the advising

agent), timing (when the agent generates advice),

intrusiveness (how proactive the agent is in

interrupting the user's workout), presentation (how

the advice is displayed to the user), and content (the

information the advice contains) (Chin-Ming Fu,

1997). Negotiation is the process by which a group

of agents come to a mutually acceptable agreement

(Jennings et al., 2001). In Figure 1, we show how

the three basic components of an E-Hub come

together and interact defining as a whole the

functionality of the system. The first component is

the Generic Product Ontology which is created so as

to cover every possible product or input

combinations which can be stored in the systems

database. The second is the Negotiation Agent, who

is responsible for managing the negotiation process

between the buyer and seller using ontology

attributes and for reaching a mutually acceptable

promise which is then fulfilled through the logistics

services. The third one is the Advising Agent, who

uses the ontology to help the user to accomplish a

specific task, keeping track of user movements and

bringing together users that share common interests

according to their profile.

A flexible and generic E-Hub architecture can

mutate from one taxonomy classification to another.

We have expanded and build upon the model

presented in (Albers et al., 1999) so as to cover

every aspect of a modern electronic hub whilst

striving to keep it as simple and hence as reusable as

possible.

Figure 1: A Generic and Agent-Mediated E-Hub Architecture

Buyer (Or group of buyers) Seller (Or group of sellers)

Advising

Agent

Product Ontology

E-Hub

Logistics

Negotiation

Agent

ICEIS 2005 - SOFTWARE AGENTS AND INTERNET COMPUTING

298

Figure 2: Generic Product Ontology

In Figure 2 the expanded product model is depicted.

The Identifier is the ID along with some details for

recognition of the product. The Physical property

corresponds to a single material when we talk about

manufacturing inputs or to a collection of raw

materials or other products so that when synthesized

an operating input is created. So both manufacturing

and operating products are supported by the

ontology (issue 1). The Functional property refers

to the possible applications of the product. The

Presentational property is related to the way in

which the product is represented to the user (Albers

et al., 1999). As described in (Niem, 1999), the latter

is accomplished by creating a 3D model from

images taken as inputs. The Product Category

property provides the vendor with the ability to

classify his product into a broader category. Each

category is assigned with specific properties called

Special Attributes. The Special Attributes property

includes alternate characteristics or meta-attributes

of a product. This property contributes to producing

a flexible system since additional product attributes

are not predefined by the ontology, but can be

created at run-time by appropriately configuring the

Product Category. In conjunction with the physical

property it provides the flexibility to the user to

promote his product or service in any way that he

sees fit (issue 1). Strategy is a property that helps

the user to define his deal-making tactics based on

the products’ negotiable attributes. Profiling is a

property that allows one to define the characters of

the people for whom the product will most likely

have greater appeal.

According to the proposed architecture buyers can

be sellers and vice versa as shown in Figure 1 by the

fact that both buyers and sellers are part of the E-

Hub. This means that an industry can buy raw

material from the same hub that it later sells the final

product. This integrates B2B and B2C E-Hubs in an

open and generic platform that everyone can

participate (issue 2). Taking into account that in

manufacturing inputs quantity determines price, the

ontology offering the Special Attributes property can

accept ranged space attributes other than price, here

quantity. In this way the E-hub supports both

vertical and horizontal business purchases (issue 1).

With the help of the Profiling attribute of the

ontology and the agent implementing it, the spot

sourcing oriented E-Hub can easily be mutated into

systematic sourcing for a specific buyer (issue 3a,

4). This, for example, can be accomplished when

the customer buys a lot from a) a specific seller that

can provide him with better terms relating, for

example, to price, and b) from not a specific seller

but from the same hub where for example having

met predefined sales levels, better prices quotes can

be offered regarding fulfilment services, etc.

The advising agent keeps track of the buyers’

movements and for the first case it informs the seller

for the specific customer and proposes him to

contract the customer with better terms. If the seller

agrees, the discount is applied every time the two

sides come to a mutually accepted agreement

through negotiation (issue 4). This mechanism

favours the vendor in the sense that he receives all

the orders and the buyer in the sense that he enjoys

better terms. If the contracts that take place consider

a discount percentage on the upper or lower limit of

the negotiation ranged space attributes and the

negotiation is still used then the buyer will still

receive lower prices. But if something unexpected

happens causing the product’s negotiable attribute

(usually price) to rise, then the seller would have a

chance to apply hard utility factors and functions on

the product that will enable him not to loose money

and to keep his client happy since he will still buy

cheaper than the others (issue 3b

For the second case, the advising agent informs the

corresponding logistics department for the discount

in shipping fees (sending them the reduced fee that

should be retrieved from his/her account) as well as

the buyer for the discount taken place. This favours

the buyers since the E-Hub lowers its

transaction/fulfilment costs. Sellers can also be

favoured in this E-Hub. The advising agent keeping

track of all the transactions within the marketplace

can also decide whether a specific seller can enjoy a

reduction in the rental space of the marketplace. This

decision is taken by considering not only the value

of the goods sold but also the frequency of sales. If

Product

Identifier

Characteristics

Physical

Components

Functional

Used For

Presentational

3D Representation

Profiling

User VS Product

Special Attributes

User-Defined attributes

Product Category

Product Belongs To

Strategy

Tactics & Scores

TRANSCENDING TAXONOMIES WITH GENERIC AND AGENT-BASED E-HUB ARCHITECTURES

299

the reduction is decided, the seller is informed and a

new contract must be signed for the changes to take

effect. So the proposed architecture could prompt us

to classify it as neutral but offering at the same time

the flexibility to become either forward, reverse and

biased (issue 4).

Last but not least, the ontology of this E-Hub

employs the ability of Reverse Aggregation since

many buyers can join together as one, to accomplish

better terms. We believe that when only reverse

aggregation is employed by an E-Hub it is unlikely

to have all buyers as possible customers. With the

architecture mentioned above, even if Reverse

Aggregation is employed, large purchasers can enjoy

better terms on their own without having to ally with

other smaller purchasers. The element client (C) is

embodied in the hub without any vitiation of the

B2B procedure (issue 2). So, everyone can

participate in this E-Hub (issue 2). This can be again

accomplished by the advising agent, who can match

sellers or buyers according to their profile and bring

them together to form a group of sellers or a group

of buyers. If this group is formed, it is then treated as

a single buyer or seller and can enjoy better terms

according to the aforementioned contracts.

4 CONCLUSIONS AND FURTHER

RESEARCH

Taxonomies are tools that can be used to classify

objects and thus, implicitly or explicitly, impose a

frame around the problem domain addressed by

research. We argue that although such framing can

be useful for guiding research by narrowing the

boundaries of the domain, they should also be

approached with caution since any such restrictions

may stifle innovation. Taking such a taxonomy as a

starting point, we have identified a set of issues that

to our opinion are obstacles to the evolution of

modern E-Hubs and proposed an architecture that

addresses them.

Our research falls under the design-science

paradigm in information systems research where

knowledge and understanding of a problem domain

and its solution are achieved by engaging in the

actual process of building the desired artifact and

applying or putting it into use.

REFERENCES

Albers, M., Jonker, C., Karami, M., Treur, J., 1999. An

Electronic Market Place: Generic Agent Models,

Ontologies and Knowledge. In Agents ‘99 Workshop

on Agent Based Decision-Support for Managing the

Internet-Enabled Supply-Chain, Seattle, Washington,

pp.71-80.

Bishop, M., Bailey, D.,1995. A Critical Analysis of

Vulnerability Taxonomies. Technical Report CSE-96-

11, University of California at Davis.

Chin-Ming Fu, M., 1997. An Architecture for

Collaborative Problem-Solving Control in Associate

Systems. Unpublished Ph.D. Dissertation, University

of Illinois at Urbana-Champaign.

Debenham, J., 2000. Supporting the actors in an

electronic market place. Last accessed September 19,

2004 at http://www-

staff.socs.uts.edu.au/~debenham/papers/ES-01.pdf

Hevner, R.A., March, T.S., Park, J., Ram, S., 2004. Design

Science in Information Systems Research. MIS

Quarterly. Vol.28, No.1, pp.75-105.

Jennings, N.R., Faratin, P., Lomuscio, A.R., Parsons, S.,

Sierra, C., Wooldridge, M., 2001. Automated

Negotiation: Prospects, Methods and Challenges.

Available at:

http://www.csc.liv.ac.uk/~mjw/pubs/gdn2001.pdf , last

accessed March 30 2004.

Kontolemakis, G., Kanellis, P., Martakos, D. 2004.

Software Agents for Electronic Marketplaces: Current

and Future Research Directions. Journal of

Information Technology Theory and Applications.

Vol.6, No.1, pp.43-61.

Mougayar, W., 2000. The Open Market Misnomer.

Business 2.0 January.

Niem, W., 1999. Automatic reconstruction of 3D objects

using a mobile camera. Image and Vision Computing.

Vol.17. No.2, pp.125-134.

Rosson P., 2000. Electronic Trading Hubs: Review and

Research Questions. Work-in-Progress Submission,

Proceedings, 16th. IMP Conference, Bath.

Kaplan, S., Mohanbir, S., 2000. E-Hubs: The New B2B

Marketplaces. Harvard Business Review, May-June,

Vol. 78, No.3, pp. 97-103.

Shen, X., Radakrishnan, T., Georganas D., 2002. vCOM:

Electronic commerce in a collaborative virtual world.

Electronic Commerce Research and Applications,

Vol.1, pp. 281-300.

ICEIS 2005 - SOFTWARE AGENTS AND INTERNET COMPUTING

300