=Paper=
{{Paper
|id=Vol-389/paper-8
|storemode=property
|title=Designing Web Content Management Systems Using the Method Association Approach
|pdfUrl=https://ceur-ws.org/Vol-389/paper08.pdf
|volume=Vol-389
|dblpUrl=https://dblp.org/rec/conf/models/LuinenburgJSWB08
}}
==Designing Web Content Management Systems Using the Method Association Approach==
https://ceur-ws.org/Vol-389/paper08.pdf
Designing Web Content Management Systems Using
the Method Association Approach
Lutzen Luinenburg1, Slinger Jansen2, Jurriaan Souer1, Inge van de Weerd2 and
Sjaak Brinkkemper2
1GX, Wijchenseweg 111, 6538 SW Nijmegen, The Netherlands
{lutzen.luinenburg, jurriaan.souer}@gxwebmanager.com
2Department of Information and Computing Sciences,
Utrecht University, The Netherlands
{s.jansen, s.brinkkemper, i.vandeweerd}@cs.uu.nl
Abstract. Model-driven web approaches focus on creating robust web
applications. There are two downsides to using these model-driven web
approaches: they consist of a unique set of models and method and they are
aimed at designing web applications from scratch. In this paper we present the
Method Association Approach, which selects and constructs fitting methods
from five model-driven web modeling approaches. The Method Association
Approach creates modeling methods in specific web application domains for
applications in different stages of their life cycle. The Method Association
Approach positions existing model-driven approaches by means of meta-model
concepts against key features of a specific web application domain. With the
Method Association Approach a design method is constructed that seamlessly
aligns with web application features, and makes use of proven web design
concepts. The approach has been validated by means of an expert validation
and two case studies.
1 Introduction
Since the beginning of web engineering [6], many web modeling methods have
proposed in order to provide efficient and structured ways of developing web
applications. However, these web modeling methods only provide design primitives
for building web applications from scratch, and not including the possibility to design
preexisting components related to a particular web domain. Web Content
Management (WCM) systems are software products used for the management of and
control of web content WCM systems have gradually merged with web applications
resulting in CMS-based web applications [26]. The implementation of WCMSs is a
complex task, since one has to cope with fast-changing requirements which have a
high impact on architecture and design. Key in this process is the translation of
business requirements into preexisting components. In order to make this process
more transparent and effective, domain-specific web design methods are needed.
Lutzen Luinenburg1, Slinger Jansen2, Jurriaan Souer1, Inge van de Weerd2 and Sjaak
Brinkkemper2
Current web modeling methods provide their own specific set of models including
a various number of design primitives. These models address design primitives which
have proven their usefulness in many cases. Therefore, it seems obvious to select
existing web modeling methods for the creation of a situational web design method
with regard to the WCM domain. Due to the high amount of web modeling methods,
each with their set of models, it is complex to make a decision on which web
modeling method to choose for applying this to the WCM domain.
In this research, we present a Method Association Approach wherein domain-
specific feature groups are associated with web modeling methods in a uniform way.
The Method Association Approach shows which feature groups are addressed by the
selected web modeling methods, and so suitable for a particular application domain.
The uniformity of the models has been established by meta-modeling. Another
advantage of meta-modeling is that it is cost-effective and efficient in comparison to
more empirical approaches (e.g. case studies).
The remainder of this paper is organized as follows. First, we present some related
work. In section two we describe the Method Association Approach. Section three
provides evaluation criteria used for method association. In section four, we show
how the association table provides insight in the coverage of feature groups by several
web modeling methods. Section five summarizes the validation of the situational web
design method by means of an expert validation and two case studies. We end this
paper with a brief discussion and conclusions.
In web engineering, many web modeling methods have been proposed in the course
of time. Many comparative studies have been performed in order to provide an
overview of these web modeling methods. Each of the comparative studies has its
own goal and provides an overview by focusing on different modeling requirements,
like: general method requirements [25], navigation modeling [8], requirements
engineering [7] and rich internet applications [21]. These comparative studies help to
get a better understanding of each of the web modeling methods, and mainly insight
in its strengths and weaknesses. However, no comparative studies have been
performed addressing modeling requirements of a particular web domain.
Hong, Van de Goor and Brinkkemper [13] provide a formal approach to the
comparison of six object-oriented methods. By creating meta-models from a process,
and data perspective as well, a uniform and formal representation of the methods is
established. From all meta-models, a super set was constructed to which each of the
methods were compared. By doing this, all similarities and differences between the
six methods were presented. In this research, also a meta-modeling technique has
been used for the uniformity of method comparison.
Besides the domain of web engineering, also the discipline of method engineering
is related to this study. Method engineering is the engineering discipline is to design,
construct and adapt methods, techniques and tools for the development of information
systems” [3]. In method engineering, a special type of engineering is used in order to
create methods that make it possible to adjust towards distinctive development project
situations at hand, called situational method engineering [28]. Situational method
engineering has been applied successfully for the assembly of situational methods
[14, 28]. A situational method can be defined as an Information Systems (IS) method
Designing Web Content Management Systems Using the Method Association Approach
that is tuned to the situation of the project at hand [12]. Ralyte et al. [22] state that an
assembly-based situational method engineering is an approach in which method
fragments are extracted and stored in a method base driven by situational method
requirements. According to Harmsen [12], a method fragment is a description of an IS
engineering method, or any coherent part thereof. Method fragments allow for
engineering situational methods in a way that it supports selection, storage and
assembly.
2 A Method Association Approach
In this section, we describe a situational method engineering approach that compares,
selects and develops web modeling methods for specific domains. Since many web
modeling methods ignore the design and development of web applications that use
preexisting features and components, we have adapted an existing assembly-based
situational method engineering approach [28]. The approach has been adapted by
adding two steps (step 2 and 5) that allow for association and selection of relevant
web modeling methods. We call this approach a Method Association Approach
(MAA).
Figure 1 gives an overview of the MAA. The approach consists of the following
seven steps:
1. Identify web engineering project situations
2. Identify feature groups of a software product
3. Select candidate methods for the identified feature groups
4. Model relevant method fragments in a method base
5. Associate feature groups with candidate method concepts
6. Assemble situational web design method
7. Validate situational web design method.
In the first step, project situations are identified by means of categorizing unique
development project characteristics. In essence, development situations allow for the
construction of situational methods . Three development project situations have been
defined, namely: standard, complex and migration development projects. Standard
development projects were defined as development projects that only need a little
customization from a software development perspective. For the standard and
complex development project situations, specific needs have been identified such as:
‘applying UML for modeling, since all developers have knowledge about this
modeling language.’ Migration project situations have not been included, since this
type of project not in the scope of this research.
During the second step, fourteen key feature groups have been identified. These
feature groups can be employed as selection and association criteria of a given
domain for the construction of situational web design methods. These feature groups
are explained in more detail in section 3.1.
Based on a literature study, four mature web modeling methods and a propriety
web design method were selected: Object-Oriented Hypermedia method (OO-H
Lutzen Luinenburg1, Slinger Jansen2, Jurriaan Souer1, Inge van de Weerd2 and Sjaak
Brinkkemper2
method) [11], UML-based Web Engineering (UWE) [15], Object-Oriented Web
Solutions (OOWS) [20], Web Modeling Language (WebML) [5] and GX-Design
method. The reason for incorporating these candidate web modeling methods were
supported by three criteria: acceptance in the community, evolution of the approach
and advanced CASE tool support. In literature, WebML, OO-H and UWE are
perceived as ‘prominent’, since these are well documented and well-supported by
design tools [29]. Also OOWS can be perceived as prominent, since it is based upon
OO-method – an automated software production environment - which has been
developed continuously since its inception in the mid 90s [19].
Existing web
Web engineering
WCMS domain modeling
projects
methods
3. Select candidate methods
1. Identify project situations 2. Identify feature
groupings
OO-H UWE OOWS
Standard
14 feature
Complex groupings GX-Design
WebML
method
Migration
5. Associate feature
groupings with
candidate methods
4. Model candidate methods
Association Method
6. Assemble situational table base
web design method
7. Validate situational
Preliminary web design method
Situational
situational
web design
web design
method
method
Fig. 1. The Method Association Approach
For modeling relevant method fragments (i.e. requirements engineering, design and
implementation method fragments), a meta-modeling technique called Process
Deliverable Diagram (PDD) was used [28]. The meta-modeling technique is based on
UML standards. For more on the modeling of candidate methods, see section 3.
In step 5, the method association and selection is performed by using the
association table. In the association table, key feature groups identified in step 2 are
associated with candidate method. This method association step is explained in more
detail in section 4.
Using the results extracted from the association table and taking in account the
development project needs identified in step 1, a preliminary situational web design
method was constructed. The situational web design method consists of four main
activities: conceptual design, architecture design, presentation design and detail
component design. All these design activities and their corresponding design
deliverables contribute to the translation of requirements into designed components.
Subsequently, these designed components will be developed and implemented in the
development phases of the implementation project.
Designing Web Content Management Systems Using the Method Association Approach
In the final step, the situational web design method has been validated by means of
an expert group and two case studies in the WCM domain. The expert validation and
the case studies have been carried out in GX, a software product company that
develops and implements a WCMS called GX WebManager [27]. With regard to the
expert validation, the situational design method was reviewed by two consultants and
two software architects. With each of the experts, we have walked through the
situational web design method and they were asked whether the method’s process and
deliverables were complete and correct. Based on the comments, the situational
design method has been adjusted. Next to that, the constructed situational web design
method has been tested in two development projects, whereof one was recently
completed [30]. In both projects a WCMS has been designed conceptually, leaving
out architecture and detailed component design.
3 Setting the Evaluation Criteria for Method Association
In this section, we discuss the evaluation criteria used for method association. First,
we elaborate on the web modeling methods used for association, and second on
feature groups that establish a web application domain.
Ceri et al. [5] state that the Web Modeling Language (WebML) is a notation for
specifying complex web sites at a conceptual level. WebML is supported by the
CASE tool WebRatio. WebML is also provisioned with a development approach
based on Boehm’s spiral model in order to build data-intensive web applications
using the notation. WebML provides modeling primitives for the conceptual level by
means of E-R schemas and, for hypertext modeling by their own propriety graphical
hypertext model. Furthermore, WebML has evolved in the course of time by
including design concepts like: workflow [2] and rich internet applications [1].
The UML-based Web Engineering (UWE) approach is an object-oriented, iterative
and incremental approach for the development of web applications [15]. The
approach focuses mainly on customized and adaptive systems. UWE uses ‘pure’
UML wherever is it possible and extends UML for navigation and presentation design
with UML extension mechanisms, also called a UML profile. UWE is supported by a
CASE tool called ArgoUWE. ArgoUWE facilitates the design process, namely:
conceptual, navigational and presentation design. Recently, UWE has been evolved
towards a Model Driven Development (MDD) approach [17].
The Object-Oriented-Hypermedia Method (OO-H) can be defined as a generic
model based on the object-oriented paradigm that captures relevant specifics related
to web application design [11]. Besides, OO-H extends OO-method with two inter-
dependent diagrams addressing navigation and presentation. OO-H is supported by a
CASE tool called VisualWade that deals with the design and implementation of web
applications. Since its inception, OO-H has been evolved by the addition of
behavioral models that deal with business processes [16] and a personalization
framework [10].
Pastor et al. [20] describe the Object Oriented Web Solutions (OOWS) method as
the extension of the software production method OO-method that introduces
expressiveness to capture the navigational and presentational requirements of Web
Lutzen Luinenburg1, Slinger Jansen2, Jurriaan Souer1, Inge van de Weerd2 and Sjaak
Brinkkemper2
applications. OOWS is supported by a commercial MDA tool called OlivaNova.
OOWS has evolved and adopted new design features, like business modeling and
semantic web technologies [9].
The GX-Design method is a design method used for the implementation of GX-
WebManager. The GX-Design method consists of three steps: architecture design,
presentation design and detail component design. In architecture design, the relative
context between the components is designed using the 4+1 view model of Kruchten
[18], and this finally results in an architecture document. During presentation design,
deliverables such as XHTML and CSS files are translated to dynamic pages. Parallel
to presentation design, customer-specific components (e.g. connectivity) are described
in a detailed component document containing design models, descriptions, and
rationale. The GX-Design method took in a special status with regard to method
association, since it is WCMS-specific and therefore different than all other methods.
In step 4 of the MA approach the web modeling methods have been modeled by
means of meta-models to provide a uniform way of representation. Meta-models have
also been applied effectively for the evaluation and comparison of methods [13].
During the process of storing all web modeling methods, a total of 32 method
fragments have been extracted and stored. From these 32 method fragments, 20
method fragments represented the main design models of the web modeling methods
and the other fragments represented abstract views on the web modeling methods.
These method fragments were extracted from main literature resources of the web
modeling methods. Besides, also available meta-models created in the MDWE project
have been used in order to speed up the modeling process. Aiming at method
integration, this project has presented meta-models of the following web modeling
methods: WebML and OO-H [29].
PRESENTATION MODEL
Analyze
navigation 1
structure
Identify navigational classes
has 1 0..*
Identify navigational relationships 0..1
MANAGERCLASS NAVIGATIONAL CLASS
0..1 1
1..2
Apply presentational 0..*
patterns 0..1 has
PAGINATION
has NAVIGATIONAL RELATIONSHIP
0..1 0..*
Define pagination patterns 0..1
INFORMATION
0..*
0..1 LAYOUT 0..1 ATTRIBUTE
Define lay-out patterns
1 has
CONTEXT RELATIONSHIP 0..1
Define ordering properties ORDERING
CONTEXT DEPENDENCY RELATIONSHIP CRITERIA
Fig. 2. A Process Deliverable Diagram of the presentational modeling phase in OOWS.
In figure 2, a PDD is presented of the presentational modeling phase in OOWS.
This phase consists of two main activities and five sub-activities, in which
presentational requirements are specified by means of patterns. In table 1 and 2, a
description of all activities and concepts is given.
Designing Web Content Management Systems Using the Method Association Approach
Table 1. Activity table for presentational modeling in OOWS.
Activity Sub-Activity Description
Analyze Identify navigational For the presentational model, the
navigation classes NAVIGATIONAL CLASSES from the navigation
structure model can be used, and therefore need to be
identified..
Identify navigational Similar to the NAVIGATIONAL CLASS, also the
relationships NAVIGATIONAL RELATIONSHIPS need to be
identified before adding presentational
requirements.
Apply Define pagination In this sub-activity, a PAGINATION-pattern that
presentational patterns specifies information "scrolling" and is applied to
patterns the MANAGER CLASS and NAVIGATIONAL
RELATIONSHIPS
Define lay-out A LAYOUT-pattern is defined that shows how
patterns content is displayed. The pattern is applicable to the
MANAGER CLASS of NAVIGATIONAL
RELATIONSHIPS.
Define ordering By defining ordering properties, a designer can
properties specify a class population ordering using the value
of one or more attributes.
Table 2. Concept table for presentational modeling in OOWS.
Concept Definition
PRESENTATION MODEL An abstract model that integrates the properties of the user
interface into the existing Conceptual Schema.
NAVIGATIONAL CLASS A class that provides a navigational view over a class in the
Class Diagram.
ATTRIBUTE A feature within a classifier that describes a range of values
that instances of the classifier may hold.
NAVIGATIONAL An unidirectional binary relationship connecting navigational
RELATIONSHIP classes.
CONTEXT DEPENDENCY A relationships that represents basic information retrieval by
RELATIONSHIP crossing a structural relationship between classes.
CONTEXT A relationship that represents basic information retrieval,
RELATIONSHIP adding a navigation capability to a target navigational
context.
MANAGERCLASS A primary navigational class of an AIU.
PAGINATION A pattern that specifies information "scrolling".
LAYOUT A pattern that specifies the layout of the information
presented.
ORDER A pattern that defines a class population ordering using the
value of one or more attributes.
3.1 Feature Groups
One added step (step 2) of the MA approach represents together with meta-models a
main ingredient that enables the comparison of web modeling methods, called feature
Lutzen Luinenburg1, Slinger Jansen2, Jurriaan Souer1, Inge van de Weerd2 and Sjaak
Brinkkemper2
groups. When developing a suitable design method for a specific web application
domain, one should have a complete overview of the design requirements of a
particular web domain. Since these design requirements that represent preexisting
components originate from business requirements these influence design. We define
these design requirements feature groups as “a class of functional design
requirements.” In short, feature groups enable comparison of existing web modeling
methods by providing means to select or find the method or web design model that
fits best with functional design requirements of a web domain.
Since no reference models are available in literature [26], feature groups have been
gathered by means of analyzing seven useful WCM market resources and a document
analysis on architecture and requirements documents, resulting in 113 features. By
means of categorization and an expert validation, the list of features has been
narrowed to a final list of 14 key feature groups.
The 14 identified feature groups are: (1) Authoring; (2) Authorization
management; (3) Community technologies; (4) Connectivity management; (5)
Content repository; (6) Deployment and replication; (7) Digital Asset Management;
(8) E-forms / transaction management, (9) Layout and presentation management;
(10); Multi-channel delivery and syndication; (11) Personalization; (12) Site
Management; (13) Web usage Mining; and (14) Workflow. These 14 identified
feature groups have been used for association with candidate method concepts, which
is explained in the following section.
4 The Association Table
In the fifth step of the MA approach (figure 1), four web modeling methods have been
associated by positioning feature groups against method fragments in a association
table. In table 3 an excerpt of the association table is given showing the association of
all identified feature groups with WebML. As identified in step 2 of the MA
approach, feature groups are listed in the right column of the association table. This
list has been made situational by means of mapping key feature groups of the WCM
domain to existing functionalities of a WCMS in particular. By making this list
situational, it is possible for any organization to create a situational web design
method that fits with the software product’s functionalities in relation to key feature
groups of a given web domain.
In this study, functionalities of the WCMS called GX WebManager are mapped
onto the feature groups. The design of a typical e-form within in WebManager is
based on the following core concepts: step, handler, validation, router and field. In
table 3, three web design models of WebML are presented including their main
concepts. The excerpt shows that a handler can be designed with an entity, operation
unit and activity in three different web design models. First, a handler represents a
kind of service or operation performed on a (data) object during a transactional
process part of e-form design in WebManager. A typical handler would be ‘create
new user in database.’ In this case, the user can be represented as an entity in the E-R
diagram of WebML. In a hypertext model of WebML, navigational requirements can
be designed, which results in an create unit that is related to an entity unit. In this
Designing Web Content Management Systems Using the Method Association Approach
create unit (subtype of operation unit) a database query on the entity can be specified
based on specific parameters. In a handler, also parameters can be initialized, queries
can be fetched from a query manager, the query can be executed resulting in a new
database record and a final redirect to a follow-up page. Alternatively, a handler can
be specified by an activity performed by a web application in the Business Process
Diagram (BPD). Additionally, WebML also provides the possibility for a designer to
create generic operational units. As depicted in table 3, only six of the fourteen key
concepts have been mapped onto WCMS functionalities, since not all feature groups
could be mapped perfectly.
Table 3. An association table representing WebML (excerpt).
WebML
E-R schema Hypertext model BPD
Interconnection sub-schema
Personalization sub-schema
Process management unit
Process sub-schema
Access sub-schema
Core sub-schema
Link parameter
Swimminglane
Operation unit
Relationship
Content unit
Display unit
User entity
Entry unit
Site view
Gateway
Selector
Activity
Entity
Page
Flow
Link
E-Form / transact. Mgmt Step x x x x x
Handler x x x
Validation x x
Router x x x x
Field x
Personalization Personalization rule x x x x x
User profile x x x x
User access x x
Site management Page x
Link x
Structure x
Media list x
Anchor x
Media overview x
Page collection x x
Connectivity management Connection external app. x x x
DB Query x x x x
Authorization User x x x x
Role x x x
Permission x x x x
Workflow Activity x x x
Status x x
Transition (flow) x x x
Authoring x
Content repository x
Deployment and replication x
Digital Asset Management x x x
Layout and presentation man.
Multi-channel delv. Sync x x
Community technologies x x x x
Web user mining x x x x
The targets to which the feature groups are positioned are the method fragments of
the four web modeling methods decomposed into three granularity levels: method,
model, and concept [4]. Some web design models have not been included, since they
have not given much attention in the main literature sources and other related
proposals. We found that relationships between the original web design models,
alternative proposals and proposals mutually are not always that clear. Next, we
Lutzen Luinenburg1, Slinger Jansen2, Jurriaan Souer1, Inge van de Weerd2 and Sjaak
Brinkkemper2
explicitly selected the main concepts of all method fragments for reasons of
conciseness.
The ‘x’ in the association table implies that a method fragment or concept is
needed for the design of a feature group or WCMS functionality. When a cell is left
‘blank’, a feature group or WCMS functionality is not addressed by any method
fragment. The notation represents two extremes and does not allow for providing
nuances and more rich information. We view the definition of a more detailed scale as
future work. Two association strategies have been applied for the qualitative analysis
on the association results being a feature group strategy and a web modeling strategy.
In other words, one can find out what feature groups are addressed by all included
web modeling methods, and the vice-versa. The purpose of the analysis is to find out
how feature groups of a particular web application are addressed by all selected web
modeling methods and more specific by its models and concepts. By performing this
analysis, a functional overview of all web modeling methods from a WCMS
perspective can be obtained.
The first association strategy used for qualitative analysis is called the ‘feature
group strategy’. In this strategy, the starting point of analysis is a feature group, and
its mapping on WCMS functionalities. From a WCMS functionality or feature group
perspective, one can find out by which web design model or concept the feature group
can be designed. Another objective of this strategy is to find differences between the
proposed design concepts of the different web modeling methods.
The second association strategy used for analysis is called the ‘web modeling’
strategy. By taking the meta-model concepts and models as a starting point, the
relative weighting of concepts and web design models part of a method is revealed.
For instance, if a concept does not address a feature group, one could wonder whether
this meta-model concept is useful. On the other hand, when a meta-model concept
addresses many feature groups in comparison to others, one could say that this
concept seems to be relatively more useful than others.
In table 4, some statistics of the three granularity levels of the web modeling
methods are given. In total, 20 models (method fragments) containing 92 concepts of
the four web modeling methods are used for association and positioned against the
key feature groups and WCMS functionalities.
Table 4. Overview of models and concepts per web modeling method.
Method Models Concepts
WebML 3 23
UWE 6 25
OO-H 4 16
OOWS 7 28
Total 20 92
In short, the association table shows how each of the feature groups is covered by
the web design models and concepts of the web modeling methods. Having
knowledge about the coverage of WCMS feature groups by the web modeling
methods, a selection of fitting method fragments can be supported. The association
Designing Web Content Management Systems Using the Method Association Approach
table provides a route map for existing web modeling methods and situational web
design methods addressing a specific domain.
4.2 From Method Association to Method Assembly
By drawing conclusions upon the association table and taking into account the
needs that come along with the different project situations, relevant method fragments
can be selected. Based on the coverage of feature groups and the requirements
specified in the development needs a situational web design method has been
assembled. The assembled situational method consists of four main activities:
conceptual design, architecture design, presentation design and detail component
design. For conceptual design, four models have been added: a domain model and
user model based on a UML class diagram, an UWE navigation model and a BPD
derived from OOWS. With regard to the quality of the situational web design method
assembly, assembly guidelines have been applied such as provided by Brinkkemper et
al. [4]. These logic rules support the assembly of a meaningful method taking in
account five method assembly criteria, such as completeness, consistency, efficiency,
applicability and reliability.
5 Method Validation
In the last step of the MA approach the situational web design method has been
validated by means of an expert group and two case studies in the domain of WCM.
The expert group consisted of two software architects and two consultants. The
experts have been selected based upon job function and experience with working
procedures. In the expert validation questions were asked about the completeness and
correctness of the methods, and more specific whether the relationships were
perceived as correct and the steps as logical precedents. Based on the expert
validation, 30 changes were requested varying from changes on a aesthetic level to
changes with high impact. The changes have led to some refinements on the
situational web design method.
After the expert validation, we have performed two case studies. One case study
organization is a large Dutch Telecommunications Provider (TP) and the other a
Dutch Governmental Organization (GO). The TP project had an estimated effort of
1500 man hours and the GO project’s effort was estimated on 1400 man hours. In
both studies, the complex route map of the conceptual design activity has been
validated. For the TP project, sixteen use cases regarding a personal user environment
have been designed from retrospective, and for the GO project, two use cases for an
intranet application were designed. After the first design, in total three involved
software developers and two consultant validated the produced models and filled-out
a survey. In this survey, for each deliverable, five questions were asked with regard to
the readability, abstraction level, correctness, tool supportability and applicability of
the conceptual design deliverables. With regard to the GO process, also questions
concerning the process in terms of structure, involvement and communication were
asked. After filling-out the survey, informal interviews were held in order to discuss
Lutzen Luinenburg1, Slinger Jansen2, Jurriaan Souer1, Inge van de Weerd2 and Sjaak
Brinkkemper2
the usefulness of all deliverables and the situational web design method as a whole,
superfluity and redundancy of information, and the availability of models in order to
support customer communication.
In order to judge the quality of case study research, one can test the validity of data
[30]. Construct validity was guaranteed by using multiple sources of evidence, which
includes interviews, surveys, requirements documents and direct observations. A
chain of evidence was created in order to establish the reliability of a case study. In
this research, a case study protocol was used which functioned as an outline for
performing the case studies and for the evaluation of the results. Next to that, all
relevant data gathered during the case studies were managed in a case study database.
From both case studies we observed that the respondents were positive concerning
the overall method, however, some differences were noted between roles and models.
First, software developers were more positive about using domain models than
consultants, since the latter considered the models too complex for customers. Both
agreed upon the usefulness for designing customized components. With regard to the
navigation model, the consultants had contradicting opinions about that. The GO
consultant perceived the navigation model as helpful in order to gain more insight in
the information structure, whereas the TP consultant questioned the supportability of
the model with regard to customer communication and the added value to design as a
whole. Besides, all respondents indicated that the user model should not be modeled
separately from the domain model. Finally, a BPD was considered as the most
readable and applicable model in comparison to other models. One software
developer stated that the BPD makes complex transaction flows of an e-form more
insightful in comparison to textual descriptions.
Fig. 3. A BPD representing an ‘edit address process’ of the TP project.
The BPD in figure 3 displays an edit address process for mobile phone subscribers
which is part of a personal user environment. After the modification of the address,
Designing Web Content Management Systems Using the Method Association Approach
the new address information is validated. When properly validated and updated in the
database, the result of the address modification is presented to the subscriber on a web
page.
6 Discussion
Applying the association table in the MA approach has some limitations. In contrast
to another formal comparison approach proposed by Hong et al. [13], no automatic
build-in validation mechanism like a ‘supermethodology’ is present, which is a target
to compare. Although, the absence of such validation mechanism does not lead to a
high subjectivity of results in the association table, because feature groups have been
described and validated, and meta-models derived from main literature resources have
been used as a formal description of the web modeling methods.
Although, the qualitative and textual description of feature groups introduces
another limitation in a way that a description does not exclude or bring insight into
overlapping feature groups. Therefore, we suggest to add a formalization step in the
MA approach with the aim to decrease the ambiguity between and of all feature
groups. This can be achieved by creating a feature group ontology for a specific web
domain.
Another limitation in the association table is that transformational relationships are
not made clearly visual and traceable in the association table. With this we mean that
it is hard to include transformational concepts multiple times in the association table,
since this will make the association table lose its clearness. Due to the lacking rules of
representing transformational relationships in the association table, we therefore
suggest to use the association table together with the meta-models in order to get a
complete overview of a particular web design model. Alternatively, another way to
solve this issue is to associate the main concepts to the feature groups that are most
related to a particular design concern (e.g. navigation, workflow, etc.).
7 Conclusion and Further Research
In this paper we present an approach for the development of situational web design
methods by means of method association. By identifying feature groups and creating
meta-models of all candidate web modeling methods, we were able to associate
feature groups of a web application domain with web modeling methods by means of
a association table. Method association is a helpful technique in order to select and
assembly method fragments that fit with domain situational factors. We have
analyzed five mature web modeling methods. The approach has been validated by
means of an expert group and two case studies.
In future research, the MA approach can be improved by means of providing
metrics for the selection of relevant methods. Rossi and Brinkkemper [23] have
proposed a set of metrics that measures the method’s complexity, and in this way a
means to method association and selection is given.
Lutzen Luinenburg1, Slinger Jansen2, Jurriaan Souer1, Inge van de Weerd2 and Sjaak
Brinkkemper2
References
1. Bozzon, A., Comai, S., Fraternali, P., and Carughi, G. T. 2006. Conceptual Modeling and
Code Generation for Rich Internet Applications. In Proc. of the 6th International
Conference on Web Engineering (ICWE 2006), Palo Alto, CA, USA. ACM, 353–360.
2. Brambilla, M., Ceri, S., Fraternali, P., & Manolescu, I. (2006). Process modeling in web
applications. ACM Trans. Softw. Eng. Methodol., 15(4), 360-409.
3. Brinkkemper, S. (1996). Method engineering: engineering of information systems
development methods and tools. Information & Software Technology, 38(4), 275-280.
4. Brinkkemper, S., Saeki, M., & Harmsen, F. (1999). Meta-modelling based assembly
techniques for situational method engineering. Inf. Syst., 24(3), 209-228.
5. Ceri, S., Fraternali, P., Bongio, A., Brambilla, M., Comai, S. & Matera, M. (2003).
Designing Data-Intensive Web Applications. Morgan Kaufman.
6. Deshpande, Y., Murugesan, S., Ginige, A., Hansen, S., Schwabe, D., Gaedke, M. & White,
B. Web Engineering. J. Web Eng., 1, 3-17 (2002).
7. Escalona, M. & Koch, N. (2004). Requirements engineering for web applications: A
comparative study. Journal on Web Engineering, 2(3), 193-212.
8. Escalona, M. J., Torres, J., Mejias, M., Guti_errez, J. J., & Villadiego, D. (2007). The
treatment of navigation in web engineering. Adv. Eng. Softw., 38(4), 267-282.
9. Fons, J., Pelechano, V., Pastor, O., Albert, M., Valderas, P., & Torres, V. (2007). Web
Engineering: Modelling and Implementing Web Applications, chapter Applying the oows
model-driven approach for developing web-applications. the internet movie data-base case-
study, (pp. 65-108). Springer London.
10. Garrigos, I., Gomez, J., Barna, P., & Houben, G.-J. (2005). A reusable personalization
model in web application design. In Proceedings of WISM'05 (pp. 40-49). Sydney,
Australia: University of Wollongong, School of IT and Computer Science.
11. Gomez, J. & Cachero, C. (2003). Information Modeling for Internet Applications, chapter
OO-H Method: extending UML to model web interfaces, (pp. 144-173). IGI Publishing:
Hershey, PA, USA.
12. Harmsen, F, Brinkkemper, S and Oei, H. (1994). Situational method engineering for
information system project approaches’, In Verrijn Stuart, A A and Olle T W (eds),
Methods and associated tools for the information systems life cycle. Proc. of the IFZP
WG8.1 Working Conference CRIS’94 Maastricht, September 1994, North-Holland,
Amsterdam, pp 169-194.
13. Hong, S., van den Goor, G., & Brinkkemper, S. (1993). A formal approach to the
comparison of object-oriented analysis and design methodologies. In Proceedings of 26th
Hawaii International Conference on System Sciences, Wailea, HI, USA (pp. 689-698).
14. Karlsson, F. & Agerfalk, P. (2004). Method configuration: adapting to situational
characteristics while creating reusable assets. Information & Software Tech., 46(9), 619-
633.
15. Koch, N. & Kraus, A. (2002). The expressive power of UML-based web engineering. In D.
Schwabe, O. Pastor, G. Rossi and L. Olsina (Eds.) Second International Workshop on Web-
oriented Software Technology (IWWOST02), pp.105–119.
16. Koch, N., Kraus, A., Cachero, C., & Meli_a, S. (2004). Integration of business processes in
web application models. J. Web Eng., 3(1), 22-49.
17. Koch, N. (2007). Classification of model transformation techniques used in UML-based
web engineering. IET Software Journal, 1(3).
18. Kruchten, P. (1995). The 4+1 view model of architecture. IEEE Softw., 12(6), 42-50.
19. Pastor, O., Insfran, E., Pelechano, V., Romero, J., & Merseguer, J. (1997). OO-method: An
OO software production environment combining conventional and formal methods. In
Designing Web Content Management Systems Using the Method Association Approach
CAiSE '97: Proceedings of the 9th International Conference on Advanced Information
Systems Engineering (pp. 145-158). London, UK: Springer-Verlag.
20. Pastor, O., Fons, J., Pelechano, V., & Abrahao, S. (2006). Web Engineering: Theory and
Practice of Metrics and Measurement for Web Development, chapter Conceptual Modeling
of Web Applications: The OOWS Approach, (pp. 277-301). Springer-Verlag Berlin.
21. Preciado, J. C., Trigueros, M. L., Sanchez, F., and Comai, S. (2005). Necessity of
methodologies to model Rich Internet Applications. In Proc. of the 7th IEEE International
Workshop on Web Site Evolution, Budapest, Hungary. 7–13.
22. Ralyte, J., Deneckere, R., & Rolland, C. (2003). Towards a generic model for situational
method engineering. Lecture Notes in Computer Science, 2681, 95-110.
23. Rossi, M., Brinkkemper, S. (1996). Complexity Metrics For Systems-Development
Methods And Techniques, Information Systems, 21, 2.
24. Siau, K., Rossi M. (1998). Evaluation of Information Modeling Methods - A Review. In
Proceedings of the 31st Annual HICCS, 5, (1998), p. 314.
25. Schwinger, W. and Koch, N. 2006. Modeling Web Applications. In Web Engineering –
Systematic Development of Web Applications, G. Kappel, B. Pröll, S. Reich, and W.
Retschitzegger, Eds. Wiley, 39–64.
26. Souer, J., van de Weerd, I., Versendaal, J., & Brinkkemper, S. (2007). Situational
requirements engineering for the development of content management system-based web
applications. Int. J. Web Eng. Technol., 3(4), 420.
27. Van Berkum, M., Brinkkemper, S., & Meyer, A. (2004). A combined runtime environment
and web-based development environment for web application engineering. In Proceedings
of the 16th International Conference on Advanced Information Systems Engineering
(CAiSE 2004), Lecture Notes in Computer Science 3084, Springer Berlin: Heidelberg, 307–
321.
28. Van de Weerd, I., Brinkkemper, S., Souer, J., & Versendaal, J. (2006). A situational
implementation method for web-based content management system-applications: method
engineering and validation in practice. Software Process: Improvement and Practice, 11(5),
521-538.
29. Wimmer, M., Schauerhuber, A., Schwinger, W., & Kargl, H. (2007). On the integration of
web modeling languages: Preliminary results and future challenges. Workshop on Model-
driven Web Engineering (MDWE), Como, Italy, July, 2007.
30. Yin, R. K. (2003). Case Study Research : Design and Methods. SAGE Publications.