TOWARDS A REPUTATION MODEL FOR ONLINE
COLLABORATIVE ERP PROJECTS
Ioannis Ignatiadis, Jonathan Briggs and Adomas Svirskas
Faculty of Computing, Information Systems and Mathematics, Kingston University, Penrhyn Road
Kingston upon Thames, SURREY KT1 2EE, UK
Keywords: ERP, reputation model, e-collaboration.
Abstract: When companies engage in online collaborations, they may need to form alliances with partners that they
have not worked with before, but who should be well placed in meeting end customer demands. Online
reputation management systems play an important role in this case in choosing the right partners to
collaborate with. Those systems facilitate the recording and dissemination of opinions of actors in an online
community, with regards to other members of that community. The purpose of this paper is to report the
research in progress carried out as part of the European Union (EU) co-funded project “PANDA”, which
examines amongst others potential reputation models that could be applied in online collaborations of Small
and Medium Enterprises (SMEs) coming together to implement an Enterprise Resource Planning (ERP)
system at an end customer.
1 INTRODUCTION
The use of reputation systems as a means to increase
trust in online communities has been examined in
the literature (e.g. Bolton, Katok, & Ockenfels,
2004; Brown & Morgan, 2006; Josang, Ismail, &
Boyd, 2007). Well-known reputation systems are
used by sites such as Amazon, eBay, Epinions, etc.
The use of reputation systems can equally well
be applied in online collaborative environments, as a
means of choosing the best partners to cooperate
with. This research is concerned in particular with
the collaboration amongst geographically dispersed
actors in the European ERP industry, comprised of
SMEs. The processes, tools and benefits of online
collaborations of such actors are examined in the EU
co-funded project “PANDA” (PANDA-Project,
2006).
The PANDA project includes the development of
an e-collaboration platform for actors (such as SME
vendors of ERP systems, their national
representatives, dealers and consultants) in the
European ERP industry. The platform provides users
with the capability to locate suitable partners across
national boundaries, form partnerships as a Virtual
Organization (VO), online manage running projects,
advertise the partners’ experiences and expertise,
and share knowledge about previously completed
projects. This paper reports the part of the PANDA
project that is concerned in particular with reputation
management to enable the selection of suitable
partners to collaborate with in the implementation of
an ERP project at an end customer.
In the sections that follow, section 2 discusses
reputation management and the envisaged approach
in the PANDA project. Section 3 presents future
research regarding reputation management in the
PANDA project. The importance of the reported
research lies in examining particular reputation
models that could be practically applicable in the
European ERP industry of SMEs, and which could
facilitate the efficient forming of strategic alliances
amongst actors in that industry.
2 REPUTATION MANAGEMENT
2.1 Reputation in Online Communities
Reputation can be defined as:
What is generally said or believed about a
person’s or thing’s character or standing.
(Josang et al., 2007)
349
Ignatiadis I., Briggs J. and Svirskas A. (2007).
TOWARDS A REPUTATION MODEL FOR ONLINE COLLABORATIVE ERP PROJECTS.
In Proceedings of the Second International Conference on e-Business, pages 349-352
DOI: 10.5220/0002107803490352
Copyright
c
 SciTePress

Online reputation systems allow members of a
community to submit their opinions (quantitative
and/or qualitative) regarding other members of the
community. This feedback is then analyzed,
aggregated and made available to the members of
that community.
Reputation mechanisms are well suited in online
marketplaces, which are characterised by a large
number of small players, often unknown to each
other, and located around the world (Dellarocas,
2003). Contractual guarantees in this case are
difficult to enforce, because of the number and
geographical spread of its participants, which also
makes repeated interactions less probable. As such,
online marketplaces rely on the reputation of its
players instead, given by other members over time,
in order to have an incentive to cooperate well with
other participants, even in a one-off deal.
2.2 Types of Reputation Systems
According to Olmedilla, Rana, Matthews, & Nejdl
(2006), the main issues with reputation systems are
the trust metrics (how to model and compute the
reputation), and the management of reputation data
(how to efficiently and securely retrieve and
compute the reputation). Regarding the last point,
Josang et al. (2007) view two types of reputation
systems: centralised and distributed. In centralised
reputation systems there is a central authority that
collects all feedback, processes it and makes it
publicly available. On the other hand, in distributed
reputation systems there are distributed stores where
ratings can be submitted, or each online participant
stores personal feedback locally, and provides this to
other parties on request, which compute the
reputation score themselves.
PANDA envisages following a mixed
(centralised/distributed) approach, as the figure
below shows. This approach is in recognition of the
fact that companies that have direct experience of
working with other companies in the past can use
their own knowledge of the quality of collaboration
with those companies, as reflected in their private
ratings. If previous experiences with relevant
companies are not sufficient to determine whether
they wish to collaborate with them or not,
aggregated ratings held centrally can be examined.
The advantage of this approach is that preference
can be given to own experiences with other partners.
In addition, as the central aggregated ratings are
computed from the local partner ratings, biases on
the aggregate ratings can be avoided.
Central
Repository
Customer
Collaboration
Partne r
Aggregate
Partner
Reputations
(Partner and
Customer-based)
Partner Ratings
of other partners
(Private knowledge)
P
r
o
v
i
d
e
P
a
r
t
n
e
r
R
a
t
i
n
g
s
P
r
o
v
i
d
e
P
a
r
t
n
e
r
R
a
t
i
n
g
s
Collaboration
Partne r
Partner Ratings
of other partners
(Private knowledge)
P
r
o
v
i
d
e
P
a
r
t
n
e
r
R
a
t
i
n
g
s
Loca l Partner
Repo s ito ry
Loca l Partner
Repository
Aggregate
Partner
Ratings
Figure 1: Reputation architecture in PANDA.
As can be seen from the figure above, partners
rate each other with regards to their collaboration in
an ERP project. They are also rated by the end
customer (individually or for the project as a whole).
The customer ratings derive from responses to a
questionnaire regarding the perceived quality of the
supplied products (the ERP system and associated
software, hardware and communications equipment)
and services. Such services can include the initial
implementation of the ERP system, upgrades,
training, support, etc. Relevant metrics to measure
customer perception of the quality of the supplied
products and services can include as applicable
(Krishnan, 1995; Stylianou & Kumar, 2000; Wu &
Wang, 2006) the relationship of the customer with
the project team, their technical expertise, the
perceived quality of the end product, its perceived
ease of use and usefulness, the success of any
required Business Process Reengineering carried
out, the level of disruption of existing business
processes, the user involvement in the project, the
overall project management by the implementation
team, the documentation and training produced, the
integration of the ERP system with other systems,
the time required for implementation, etc.
The partner ratings for each other in the
collaborative ERP project can then include as
applicable the level of collaboration amongst the
partners, the perceived business and technical
expertise of the other partners, their business and
cultural awareness of the context where the ERP was
implemented, their leadership skills, their adherence
to time schedules, etc.
ICE-B 2007 - International Conference on e-Business
350

2.3 Reputation Dimensions
Following the distinction of a reputation model into
its individual, social and ontological dimensions
(Sabater & Sierra, 2001, 2002), the following figure
illustrates its relation to PANDA. The individual
dimension consists of ratings that each company
holds locally with regards to its own perception of
other companies (corresponding to the distributed
type of reputation). The social dimension consists of
an aggregation of all such relevant local ratings
(corresponding to the centralised type of reputation).
Both of the individual and social dimensions are
further differentiated into the ontological dimension,
which can include the type of service provided (e.g.
ERP installation, customization, upgrade, training,
support, etc), the particular ERP/module concerned,
the industry where the ERP was implemented, and
the geographical region. Each of these dimensions
can then be further subdivided into lower levels of
detail.
Individual
Dimension
Social
Dimension
Ontological
Dimension
Central
Repository
Loc al
Partner
Repository
- Type of Service
- ERP Module
- Industry Vertical
- Geographical Region
Aggregate
Partner &
Customer
Ratings
Partne r
Customer
Own Ratings of Other Partners Ratings of Project / Partners
Figure 2: Reputation dimensions in PANDA.
2.4 The Role of Agents
Software agents are foreseen to aid in the
management of reputation in PANDA, mostly when
a critical mass of companies is involved. The agents
in this case can aid in automatically determining the
ratings of other partners. As before, the locally
stored ratings can take precedence in determining
the rating of a potential partner. If those are not
enough to confidently determine partner reputation,
then other agents can be asked to give their own
ratings of the particular partner. This approach then
gives rise to questions of the reliability of those
ratings, the trustworthiness of the relevant agents,
the similarity with own ratings, and the potential
confidentiality of other ratings (Huynh, Jennings, &
Shadbolt, 2006; Mui, Halberstadt, & Mohtashemi,
2003; Yu & Singh, 2002). If this approach doesn’t
yield any satisfactory results, then as before the
central repository holding aggregate ratings can be
examined.
Local Partner
Repository
Partner
Software
Agent
Reliability
Trustworthiness
Closeness of Ratings
Confidentiality
Software
Agent
Partner
Central
Repository
Customer
Aggregate
Partner &
Customer
Ratings
Provide Ratings Provide Ratings
Provide
Ratings
Local Partner
Repository
Figure 3: Role of agents in reputation management in
PANDA.
3 FUTURE WORK
Regarding the engine or method of computation,
Josang et al. (2007) cite some indicative models
such as simple summation or average of ratings,
bayesian systems, discrete trust models, belief
models, fuzzy models and flow models.
The simplest one is to sum the number of
(positive or negative) ratings and keep a total score.
The advantage of this approach is that it is very easy
for anyone to understand the approach behind the
calculation; the disadvantage is that it is primitive
and can give a poor picture of an entity’s reputation
score. A slightly more advanced scheme would be to
compute the average of ratings, and a further
refinement would be to compute a weighted average
TOWARDS A REPUTATION MODEL FOR ONLINE COLLABORATIVE ERP PROJECTS
351

of all ratings in order to determine partner
reputation.
As a requirement for PANDA is to have an
approach which is simple to understand by its users,
a weighted average is envisaged to be used. This
however brings in the question of the way that
different weights are calculated and used in the
reputation of partners according to ERP project
particularities. This includes determining the
importance and consequent weight of each project
partner, the weights of different measurements items
(as indicated in section 2.2), the weights of ratings
given by the customer and the other partners, as well
as the weights given to the ontological dimensions
of a project (e.g. according to the type of service
provided, ERP module, industry vertical and
geographical region). Determining those weights
and their combination is a complex process, which
can be aided with the use of software agents once
deployed.
When using software agents, the distributed
computation of reputation would also include asking
other agents for their ratings of a partner. This would
then mean that the weights given to the ratings of
other agents would have to be determined according
to the trustworthiness and existing reputation of
those agents, the age of their ratings, the distance
between their ratings and the existing partner rating,
etc. Although such aspects of agent behaviour have
been addressed in the literature (e.g. Huynh et al.,
2006; Mui et al., 2003; Yu & Singh, 2002), their
practical applicability and business acceptance in an
online environment is an interesting research
problem to examine. To aid in initial tests of agent
behaviour, simulations will be carried out, and user
feedback (from European actors in the ERP
industry) elicited. Amongst others, the research
agenda also includes whether examining group
reputations instead of individual reputations yields
more trust in online collaborative ERP projects, e.g.
by determining which types of partners work best
with each other as a team as opposed to examining
binary pairs.
Although the PANDA project is exemplified in
the European ERP industry for SMEs, the currently
researched reputation model could also be applied in
other settings where online collaborative projects are
implemented. This includes practically any business
sector where business-oriented software solutions
(i.e. software products coupled with value added
services to form ‘extended’ solutions) are used. As
such, the PANDA project is important in serving as
a demonstrator and proof-of-concept for future
research and development in the area of online
collaborative environments.
REFERENCES
Bolton, G. E., Katok, E., Ockenfels, A., 2004. How
Effective Are Electronic Reputation Mechanisms? An
Experimental Investigation. Management Science,
50(11), 1587-1602.
Brown, J., Morgan, J., 2006. Reputation in Online
Auctions: The Market for Trust. California
Management Review, 49(1), 61-81.
Dellarocas, C., 2003. The digitization of word-of-mouth:
Promise and challenges of online feedback
mechanisms. Management Science, 49(10), 1407–1424.
Huynh, T. D., Jennings, N. R., Shadbolt, N. R., 2006. An
integrated trust and reputation model for open multi-
agent systems. Journal of Autonomous Agents and
Multi-Agent Systems, 13(2), 119-154.
Josang, A., Ismail, R., Boyd, C., 2007. A survey of trust
and reputation systems for online service provision.
Decision Support Systems, 43(2), 618-644.
Krishnan, M. S., 1995. Determinants of service quality in
software: an empirical analysis. In 1995 conference of
the Centre for Advanced Studies on Collaborative
research. Toronto, Ontario, Canada.
Mui, L., Halberstadt, A., Mohtashemi, M., 2003.
Evaluating Reputation in Multi-agents Systems. In R.
Falcone, S. Barber, L. Korba & M. P. Singh (Eds.),
Trust, Reputation, and Security: Theories and
Practices (pp. 123-137). Berlin. Springer-Verlag.
Olmedilla, D., Rana, O. F., Matthews, B., Nejdl, W., 2006.
Security and Trust Issues in Semantic Grids. In
Dagstuhl Seminar Proceedings. Internationales
Begegnungs- und Forschungszentrum fuer Informatik
(IBFI).
PANDA-Project. (2006). EU IST-027169. Retrieved May,
2007, from http://www.panda-project.com
Sabater, J., Sierra, C., 2001. REGRET: A reputation
model for gregarious societies. In 4th Workshop on
Deception, Fraud and Trust in Agent Societies.
Montreal, Canada.
Sabater, J., Sierra, C., 2002. Reputation and Social
Network Analysis in Multi-Agent Systems. In
AAMAS'02. Bologna, Italy, July 15-19.
Stylianou, A. C., Kumar, R. L., 2000. An Integrative
Framework for IS Quality Management.
Communications of the ACM, 43(9), 99-104.
Wu, J.-H., Wang, Y.-M., 2006. Measuring ERP success:
the ultimate users' view. International Journal of
Operations & Production Management, 26(8), 882-903.
Yu, B., Singh, M. P., 2002. Distributed Reputation
Management for Electronic Commerce.
Computational Intelligence, 18(4), 535-549.
ICE-B 2007 - International Conference on e-Business
352