MATCHING ERP FUNCTIONALITIES WITH THE LOGISTIC
REQUIREMENTS OF FRENCH RAILWAYS
A Similarity Approach
Iyad Zoukar , Camille Salinesi
Centre de Recherche en Informatique, Université Paris 1 – Panthéon – Sorbonne
90, rue de Tolbiac 75013 Paris – France
Keywords: ERP, Business Process, Requirements Matching, Meta-Modelling, Similarity Analysis, Requirements
Elicitation.
Abstract: Ensuring the adequacy of Enterprise Resource Planning (ERP) implementations to business requirements is
still an issue that needs to be addressed. One important cause of inadequacy results from the lack of
attention paid to the precise and systematic analysis of how well ERP functionalities match the business
requirements. One well known reason for this is that the language used to define ERPs is different from the
one used to define business requirements, and there is no technique available so far to evaluate
systematically similarities between ERP functionality models and business requirements models. Our
approach to this issue is (i) to materialise with a unified goal/strategy modelling language both the ERP
functionalities and the business requirements, and (ii) to systematically specify using a similarity model how
a given ERP functionality model and a business requirement model should match together. This paper
outlines this matching method and explains how the similarity model was developed in a systematic way.
1 INTRODUCTION
Ensuring the adequacy between an organisation’s
requirements and the Information System (IS)
functionalities represents a major issue of any IS
project. This especially holds with ERP systems in
which functionalities are already designed and built-
in for standard business processes. Our work at
SNCF (the French railways company) consists in
developing a methodology that would help specify
how the functionalities provided by the PeopleSoft
ERP match the requirements of the supply chain
processes that it shall support.
As often in this kind of project, the approach
in
itially taken in the project was mostly driven by
PeopleSoft functionalities. However, these are only
documented at a very detailed level transactions,
operations and data structures, etc), whereas at
SNCF stakeholders think in terms of their goals,
tasks and outcome. This results in a difficulty for
stakeholders to adapt to the language of ERP
experts, and furthermore, in a difficulty to foresee
how the system will fit to their requirements. This
mismatch exposes the project to a classical and well-
documented danger of failure (Standish, 1995)
(Davenport, 2000).
Our approach was that the relationship between the
sy
stem and the business should be materialised using
a unique language that gathers the business and the
system perspectives (Salinesi, 2003). The formalism
we use, called Map (Rolland, 2001), is built on two
central concepts that are natural to business experts,
goals and strategies: goals are used to identify the
business processes for which the system provides (at
least partial) support; strategies indicate how the
business intends to achieve goals.
Our purpose at SNCF was to establish a way of
work
ing to align the business requirements and
system functionalities. To do that, we adapted the
general IS evolution framework (Salinesi, 2003).
The framework that we developed clearly shows that
an important aspect of matching is the production of
statements about the similarity between the ERP and
the business requirements. A typology of similarity
predicates was thus developed.
The next section introduces the matching
fram
ework. In section 3 the similarity topology is
presented. Examples from the SNCF project are
given in section 4. Related works and conclusions
are respectively presented in sections 5 and 6.
444
Zoukar I. and Salinesi C. (2004).
MATCHING ERP FUNCTIONALITIES WITH THE LOGISTIC REQUIREMENTS OF FRENCH RAILWAYS - A Similarity Approach.
In Proceedings of the Sixth International Conference on Enterprise Information Systems, pages 444-450
DOI: 10.5220/0002628404440450
Copyright
c
SciTePress
Might-Be
As-WishedAs-Is
BM BM
To-Be
BM
SFM
To-Be
SFM
SNCF
Formalism
ERP
Formalism
Map
Map
Map
Similarity
Matched
Map
Fitness Fitness
uses
Evaluate
Matching
Decide
Matching
Matching Process
Might-Be
As-WishedAs-Is
BM BM
To-Be
BM
SFM
To-Be
SFM
SNCF
Formalism
ERP
Formalism
Map
Map
Map
Similarity
Matched
Map
Fitness Fitness
uses
Evaluate
Matching
Decide
Matching
Matching Process
Figure 1: Adapting the Evolution Methodological Framework to the context of the ERP project at SNCF
2 ERP MATCHING FRAMEWORK
As proposed by (Jarke, 1993), a general framework
was defined to situate the different concepts that are
needed in a method dealing with IS evolution.
Besides, there are different contexts of IS evolution;
each context can be characterised with a more
specific framework (Salinesi, 2003).
One of the evolution contexts concerns IS
customisation from a product family. This context is
met in ERP implementation which corresponds to
the SNCF project nature. In the generic
customisation from a product family’ framework,
four families of models have to be managed. The
requirements of the organisation are expressed in the
As-Wished BM (Business Model). The Might-Be
SFM (System Functionality Model) reflects the
functional capability of the product family. The To-
Be BM and its counterpart, the To-Be SFM result
from a matching process which searches for the best
fit between the organisational requirements
(expressed in As-Wished BM) and what is proposed
by the ERP, (Might-Be SFM).
We adapted this generic framework in order to take
into account all the kinds of models actually used in
the SNCF project. As shown in figure 1, numerous
kinds of models were used. These are organised in
the four aforementioned families: (1) As-Wished BM
capture the business processes that the organisation
requires for the future. (2) The Might-Be SFM
represent the different functionalities provided by
the ERP. (3) The To-Be BM represent the business
processes after the project. (4) The To-Be SFM
specify the ERP after the project, i.e. after
parameterisation, re-development of existing
functions, and development of new specific
functions.
An important goal of the Matching Process is to
ensure the fitness relationship between the business
and the system at the end of the project, i.e. to make
sure that the delivered system fits to the use of the
future business. On the business level, this calls for
adaptation of the Business Processes; on the system
functionality level, this calls for customisation of the
ERP and of the legacy system.
As illustrated above, the situation at SNCF is a
complex one as different formalisms are used in
each of the four families of models that does not
allow a systematic matching. Besides, the project
managers found that the business processes are too
complex to be defined at once, and therefore chose
to develop them by analysing the current situation.
This choice appears through a fifth family of
models, namely the As-Is BM.
Given this complex situation, our proposal was to
use the map formalism to reflect the existing As-Is
BM, As-Wished BM and Might-Be SFM in a unique
way. Let us notice that, the other way round, the
additional work that was needed to develop these
maps was also useful as it allowed to synthesise the
models at hand in abstract terms and remove
cumbersome details. Last, the matching process was
facilitated as it resulted in producing similarities
between specifications expressed with a unique
meta-model.
3 THE SIMILARITY TYPOLOGY
The central technique used during the matching of
Business Models and System Functionality Models
was the analysis of similarities between them.
Intuitively, a similarity (represented with the “
symbol in Figure 2) expresses a resemblance
between elements from two different models. Our
assumption is that similarities between maps can be
expressed with predicates that can be listed under
the form of a structured typology.
MATCHING ERP FUNCTIONALITIES WITH THE LOGISTIC REQUIREMENTS OF FRENCH RAILWAYS: A
SIMILARITY APPROACH
445
3.1 Towards a generic similarity
typology
Obviously a relevant map similarity typology could
have been defined in an ad-hoc sort of manner.
However, besides the fact that this might be error
prone, the resulting typology would be dependent on
the specific requirements specification formalism
used in this paper, i.e. the map formalism. To
overcome these difficulties, we adopted the
approach developed in (Rolland, 2003) (Etien, 2003)
to specify a typology of gaps between maps. First,
we searched for a generic similarity typology, i.e. a
typology that is independent of the formalism used
to express the As-Wished and the Might-Be models.
Then, the map similarity typology was developed as
an instance of the generic similarity typology. The
instantiation was guided by an accurate definition of
the map models in terms consistent with those of the
generic similarity typology. Using this approach,
other similarity typologies such as a similarity
typology between object oriented models or a Use
Case could be easily generated too.
As shown in figure 2, a generic level is first used to
abstract the specific meta models used in ERP
projects. The left part of figure 2 shows that the
generic meta-model is instantiated by specific meta-
models, and that specific similarity typologies are
issued from the generic similarity typology. The
purpose of the former instantiation is to identify the
key elements and structures of the specific meta-
model. The definition of the specific similarity
typology is made systematic by the knowledge of
the link between the similarity typology and meta-
model at the generic level, combined with the
instantiation relationship between the generic meta-
model and the specific meta-model. A description of
the generic Meta-Model can be found in (Etien,
2003).
3.2 The generic similarity typology
Any meta model is composed of elements with
properties. Besides, the structure of meta models is
shown though element composition and through
links between elements. Based on this, the generic
typology of similarity predicates emphasises that
given a pair of elements, (i) their properties can be
similar, and (ii) their structure can be similar. As the
right part of figure 2 shows, there are thus two
classes of similarities, intrinsic similarity and
structural similarity.
A pair of elements has an intrinsic similarity if they
have similar properties. Element properties can be
considered similar if they have a close semantics. In
the first place, intrinsic similarity relates to
synonymy. However, it can also be about the
hyponymy (or the other way round hyperonymy)
relationship.
The structural similarity deals with the composition
of elements and their organisation within models.
There are thus two classes of structural similarity:
compositional similarity, and relational similarity.
Contrary to intrinsic similarity that involves only the
two compared elements, the structural similarities
imply comparisons between other elements that are
related to the two compared ones.
Besides, as shown in figure 2 by the aggregation link
from structural similarity class to the similarity
class, a structural similarity is a complex one and
involves other similarities. For example two
elements have the “same components in a
composition” if each component in one element has
a semantically “same” counterpart in the
composition of the other element.
There are therefore four main similarity classes.
These are defined as follows:
(i) Synonymy is a relationship between two elements
that have either properties or types with a similar or
close meaning. There are two kinds of synonymy:
- Two elements have a synonym type if their types
are equal or have a common super-type (they are
then cousins).
- There are different degrees of resemblance possible
between the properties of a pair of elements: two
elements have the same property when their
properties have exactly the same name and the same
Issued from
Typology of
Similarity
Applies on
elements of
Generic
Meta-model
Issued from
Generic
Meta-model
level
Generic Similarity
Typology
Applies on
elements of
Model
level
Specific
Meta-model
level
Instance of
Instance of
As-Wished
Model
Might-Be
Model
Instance of
Meta
Model
Issued from
Typology of
Similarity
Applies on
elements of
Generic
Meta-model
Issued from
Generic
Meta-model
level
Generic Similarity
Typology
Applies on
elements of
Model
level
Specific
Meta-model
level
Instance of
Instance of
As-Wished
Model
Might-Be
Model
Instance of
Meta
Model
Similarity
Intrinsic Structural
Hyperonymy/
Hyponymy
Synonymy
Compositional
Relational
The generic similarity typology
Figure 2: Overview of the similarity elicitation approach
ICEIS 2004 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION
446
meaning (for example an actor in a Use Case model
and an actor in a sequence model); they have alike
properties when their properties are identified with
different words but have the same meaning (for
example two classes that specify the same business
object in two different ERP modules); or they have a
resembling property when the properties have
different names and values, but they still have a
close meaning (like for example a standard business
object in two different ERPs).
(ii) Hyponymy/Hyperonymy relates two elements
when the meaning of the one subsumes/is subsumed
by the meaning of the other. As with synonymy,
hyponymy/hyperonymy similarity can be defined on
the type and on the properties of elements:
- with respect to type, hyponymy/hyperonymy
comes down to a father or a son relationship
between the types of the involved elements. This is
for instance the case of a UML class in a model that
appears as an abstract class in another model.
- With respect to properties, two elements are in a
hyponymy/hyperonymy relationship if the properties
of the ones includes/extends the properties of the
other. This is for instance the case when the
attributes of one class are included in the collection
of attributes of another class.
(iii) Relational similarities are defined between link
elements that are connected to similar source/targets,
or between elements that are related to the rest of
their models through similar links. As table 1 shows,
there are different kinds of relational structure
similarity predicates.
These include (without being restricted to): same
number of links (when two elements are
source/target of the same number of links), same
number of links entering in a node (when two
elements are source of the same number of links),
same number of links outgoing from a node (idem,
the other way round), same/alike/resembling source,
target, or source and target (when two links have
similar extremities), same depth (same max distance
between nodes and leaves of the trees they belong
to) or same height (same max distance between
nodes and the root of the trees they belong to).
(iv) Compositional similarities deal with compound
elements that are similar in their composition, and
with elements that belong to similar compositions.
Table 1 quotes a number of compositional structure
similarity predicates: same cardinality of a
component (when two compound elements have the
same number of components), same / alike /
resembling components in a composition (when the
compositions of two compound elements are
comparable), same / alike / resembling common
component in a composition (when part of the
compositions are comparable), part of same / alike /
resembling compound (when an element is similar to
a component of another element).
4 EXAMPLES OF APPLICATION
AT SNCF
Three sets of maps were produced as required in the
framework presented in section 2: As-Is, As-Wished
and Might-Be maps respectively representing the
current BP at SNCF, the wished BP at SNCF, and
the BP supported by PeopleSoft. This section reports
examples of use of similarity typology to match
those maps in order to produce the matched maps
that represent the To-Be, i.e. the business and the
system situations after the ERP implementation.
In figure 3 the two extracts of maps represent the
organisation requirements (on the right) and what is
proposed by PeopleSoft (on the left). A number of
similarities can be specified using the typology of
similarity predicates. For example, there are intrinsic
similarities between the As-Wished and Might-Be
maps. Indeed, synonymies with equal types and
equal properties can be defined between the pair of
intentions Build production plan and Solve
production plan and between the two strategies For
items managed with the forecast and For items
managed with orders.
Although their statement
looks very different at first glance, there is a
hyponym/hyperonym similarity between the
Table 1: The generic similarity predicates
Synonymy
Hyperonymy
Hyponymy
Relational Compositional
Type
Type Same number of links Same cardinality of a component
Equal type Father type Same links number entering in a node Same components in a composition
Cousin type Son type Same links number outgoing from a node Alike components in a composition
Same source Resembling components in a composition
Property
Property Alike source Same common component in a composition
Same property Includes property Resembling source Alike common component in a composition
Alike property Extends property Same target Resembling common components in a composition
Resembling property Alike target Part of same compound
Resembling target Part of alike compound
Same source & target Part of resembling compound
Alike source & target
Resembling source & target
Same depth
Same height
MATCHING ERP FUNCTIONALITIES WITH THE LOGISTIC REQUIREMENTS OF FRENCH RAILWAYS: A
SIMILARITY APPROACH
447
By frozen fence and the Based on the first two
months of the planning horizon strategies as the
former proposes a more general solution with more
options than the latter. Besides, looking at the
relational similarities between the two map extracts,
it was found that they have the same depth. This
indicates that they are described at the same level of
abstraction. A closer analysis of the bundles
composed of the pair of goals and the strategies
between them shows that there are however
compositional similarities. As one could expect from
the different number of strategies, the two bundles
do not have the same components. However, the
bundles have an alike common components
similarity as two of the strategies in the As-wished
bundle have an equal in the Might-Be, and one has a
hyperonym in the Might Be. No similarity could be
found by looking at the height with respect to the
refinement links (these links are not shown in the
figure for the sake of space).
This experience showed us that a number of
alternative To-Be solutions could always be
envisaged. This calls for a systematic way to explore
alternatives, i.e. to identify them and to find the best
To-Be solution. Map similarity typology facilitates
the identification and the construction of the
different matching alternatives. Because similarity
predicates are generic, automation of similarities
detection is possible using our generic typology. In
real project size, the number of As-Wished and
Might-Be models can by very great and manual
similarities detection is impossible that why CASE
tools are needed.
Similarities discovered using the generic typology
present an important element of a decision making
process during the matching process to choose the
best solution for the organisation.
Figure 4 shows an example of situation in which
alternative matchings can be explored. In this
situation a number of wished business strategies and
system functionalities were identified to ‘Solve
production plan’. These are respectively shown by
the right and left map extracts in Figure 3. Matching
these maps using the above leads to at least three
possible To-Be maps as Figure 4 shows it:
Alternative A1 results from an As-Wished driven
matching, i.e. the solution is mostly similar to the
As-Wished BM. Alternative A2 is issued by a
Might-be driven matching. This solution adopts all
the strategies from the Might-Be SFM that can be
useful to SNCF. Alternative A3 can be surfaced by a
Two-way matching.
Discussions were raised for choosing among the
alternatives resulting from the matching process.
Decisions could be made using a number of criteria
such as the percentage of specific development in
the ERP, non-regression, cost, delay, available
competencies, negotiation margins, etc.
Using
pegged
chains
Using
planning
solvers
Using plan summary
Using exception inquiry
Using details inquiry
Revison strategy
By balancing multiple
business constraints
simultaneously
Using
operational
exceptions
By frozen
fence
For items
managed
with orders
Solve production plan
For items
maneged
with the
forecast
By consuming inventory
By using resources
By supplying inventory
Build production plan
Extract from Might-Be Maps
Based on the first
2 months of the
planning horizon
For items managed
with orders
Solve production plan
For items maneged
with the forecast
Build production plan
Extract from As-Wished Maps
Figure 3: Two extracts from As-Wished and Might-Be maps
Common strategies in the As-Wihed
Maps and Might-Be Maps
Strategies selected from Might-Be Maps and which do not
have equivalents in As-Wished Maps
Using
pegged
chains
Using
planning
solvers
Revison strategy
By balancing
multiple business
Constraints
simultaneously
Using
operational
exceptions
By frozen
fence
For items
managed
with orders
Solve production plan
For items
maneged
with the
forecast
Build production plan
A2
By frozen
fence
Solve production plan
For items
maneged
with the
forecast
Build production plan
For items
managed
with orders
A1
Using
pegged
chains
Using
planning
solvers
Revison strategy
By frozen
fence
Solve production plan
For items
maneged
with the
forecast
Build production plan
For items
managed
with orders
A3
Common strategies in the As-Wihed
Maps and Might-Be Maps
Strategies selected from Might-Be Maps and which do not
have equivalents in As-Wished Maps
Common strategies in the As-Wihed
Maps and Might-Be Maps
Strategies selected from Might-Be Maps and which do not
have equivalents in As-Wished Maps
Using
pegged
chains
Using
planning
solvers
Revison strategy
By balancing
multiple business
Constraints
simultaneously
Using
operational
exceptions
By frozen
fence
For items
managed
with orders
Solve production plan
For items
maneged
with the
forecast
Build production plan
A2
Using
pegged
chains
Using
planning
solvers
Revison strategy
By balancing
multiple business
Constraints
simultaneously
Using
operational
exceptions
By frozen
fence
For items
managed
with orders
Solve production plan
For items
maneged
with the
forecast
Build production plan
Using
pegged
chains
Using
planning
solvers
Revison strategy
By balancing
multiple business
Constraints
simultaneously
Using
operational
exceptions
By frozen
fence
For items
managed
with orders
Solve production plan
For items
maneged
with the
forecast
Build production plan
A2
By frozen
fence
Solve production plan
For items
maneged
with the
forecast
Build production plan
For items
managed
with orders
A1
By frozen
fence
Solve production plan
For items
maneged
with the
forecast
Build production plan
For items
managed
with orders
By frozen
fence
Solve production plan
For items
maneged
with the
forecast
Build production plan
For items
managed
with orders
A1
Using
pegged
chains
Using
planning
solvers
Revison strategy
By frozen
fence
Solve production plan
For items
maneged
with the
forecast
Build production plan
For items
managed
with orders
A3
Using
pegged
chains
Using
planning
solvers
Revison strategy
By frozen
fence
Solve production plan
For items
maneged
with the
forecast
Build production plan
For items
managed
with orders
Using
pegged
chains
Using
planning
solvers
Revison strategy
By frozen
fence
Solve production plan
For items
maneged
with the
forecast
Build production plan
For items
managed
with orders
A3
Figure 4: Alternative matching possibilities
ICEIS 2004 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION
448
5 RELATED WORKS
Despite the widespread adoption of ERPs by
business organizations (Joseph, 1998), there is still a
lot of academic research need on ERPs from the
engineering perspective (Borell, 2000).
Two families of approaches were developed to
address the matching issue: the management
approaches and the system approaches. In the
management family, research intends to define the
impact of ERP installation on corporate culture
(Krumbholz, 2000), organization (Robey, 2002), or
business processes (Esteves, 2002). In the system
family, the purpose has been to guide the
identification and selection of the most appropriate
ERP (Ncube, 2000), and to elicit requirements to
inform the most adequate customisation of ERPs
(Finkelstein, 2002).
Our approach is in-between the two families. Its
main assumption is that in an ERP project, the issue
of organizational change and system engineering are
intertwined. Therefore, their matching should be
neither driven by the business nor driven by the
system functionality, but both.
Several works have already been achieved on
similarity measure. For example, (Castano, 1993)
proposes to evaluate components reusability through
conceptual schema. (Jilani, 1997) used similarity
measures to select best-fit components. Similarity
metrics for heterogeneous database schema analysis
were introduced by (Bianco, 1999). Our similarity
approach is inspired by Castano and Bianco.
6 CONCLUSION
Our experience in an ERP project at SNCF
confirmed to us the importance of the matching
between the world of business and the one of
systems and the necessity of using similarity
analysis techniques to support this matching. Our
approach to this issue was to use a goal/strategy
model called map as an in-between language on
which the matching process can be achieved in an
efficient way. So far, we have: (i) developed a
specific methodological framework that sets in
context the business models, system functionality
models, and the matching approach, (ii) defined the
issues of matching business models and system
functionality models and (iii) adopted a similarity
typology to systematise the specification of the
result from matching activities.
The next tasks in our research program are: to
further document our methodological framework, to
complete the similarity typology and to develop a
completely guided methodological process model.
REFERENCES
Bianco G. and al. A Markov Random Field Approach for
Querying and Reconciling Heterogeneous Databases.
DEXA’99, Florence, 1999.
Borell A. Hedman J. CVA Based Framework for ERP
Requirement Specification. IRIS 23. Laboratorium for
Interaction Technology, University of Trollhattan
Uddevalla, 2000.
Castano S. De Antonellis V. A Constructive Approach to
Reuse of Conceptual Components. 2
d
International
Workshop on Software Reusability., Italy, 1993.
Davenport T.H. Mission Critical: Realizing the Promise of
Enterprise Systems. Harvard Business. School Press,
MA, 2000.
Esteves J. Pastor J. Casanovas J. Monitoring Business
Process Redesign in ERP Implementation Projects.
Americas Conference on IS, Dallas, 2002.
Etien A. Salinesi C. Towards a Systematic Definition of
Requirements for Software Evolution: A Case-study
Driven Investigation, EMMSAD Workshop in the
CAISE Conference, Austria, 2003.
Finkelstein A. Bush D. Requirements Elicitation for
Complex Safety Related Systems. London
Communication Symposium. London, 2002
Jarke M. Pohl K. Establishing visions in context: toward a
model of requirements process. 12
th
International
Conference Information System. Orlando, 1993.
Jilani L.L. Mili R. A. Mili. Approximate Component
Retrieval: An Academic Exercise or a Practical
Concern?. WISR8, Columbus, Ohio, 1997.
Joseph T. Swanson E.B. The Package Alternative in
Systems Replacement: Evidence for Innovation
Convergence, in Information Systems Innovation and
Diffusion. Larsen-McGuire (eds.), 1998.
Krumbholz M. Maiden N.. How Culture might impact on
the Implementation of Enterprise Resource Planning
Packages. CAISE'00, Springer Verlag, 2000.
Ncube C. Maiden N. COTS Software Selection: The Need
to Make Tradeoffs Between System Requirements,
Architectures and COTS/Components. COTS’00
Workshop for ACSE00, Ireland, 2000.
Robey D. Ross J.W. Boudreau M-C. Learning to
Implement Enterprise Systems: An Exploratory Study
of the Dialectics of Change. Journal of Management
Information Systems, 2002
Rolland C. Prakash N. Matching ERP System
Functionality to Customer Requirements. 5
th
IEEE
Requirements Engineering, Toronto, 2001.
Rolland C. Salinesi C. Etien A. Eliciting Gaps in
Requirements Change. Requirements Engineering
Journal, 2003.
MATCHING ERP FUNCTIONALITIES WITH THE LOGISTIC REQUIREMENTS OF FRENCH RAILWAYS: A
SIMILARITY APPROACH
449
Salinesi C. Rolland C. Fitting Business Models to
Software Functionality: Exploring the Fitness
Relationship. CAISE'03, Springer Verlag (pub), 2003.
Standish. Standish
Group. Chaos. www.standishgroup.com
/sample_research/chaos_1994_1.php, 1995.
ICEIS 2004 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION
450