Content uploaded by Erki Eessaar
Author content
All content in this area was uploaded by Erki Eessaar on Feb 23, 2014
Content may be subject to copyright.
Integrated System Analysis Environment
for the Continuous Consistency and
Completeness Checking
Erki EESSAAR
Department of Informatics, Tallinn University of Technology, Raja 15, 12618 Tallinn,
ESTONIA
eessaar@staff.ttu.ee
Abstract. Modelers create models of a system during information system
development. Models should be correct, complete and consistent in order to be
most useful. Modeler must have possibility to use software that helps to automate
these kinds of checks in order to raise the speed of work and quality of its results.
This article describes the software system that allows to perform strategic- and
detailed analysis of a database-centric information system without using
complicated visual notations. System supports methodological framework for the
Enterprise Information System (EIS) strategic analysis (like the one that is
described by Roost et al. [1]). User of the proposed system records specification in
a database as a single integrated model. We propose database queries that help to
find consistency and completeness (CC) problems of the recorded models.
Keywords. Model multiplicity, metamodeling, consistency, completeness, CASE
Introduction
Successful development of an information system takes a lot of effort. Important part
of this work is modeling of the system. Common approach is to create different types
of models that describe different aspects (dimensions) of the system with the different
level of abstraction. An example is the Zachman framework for the information
systems architecture [2]. Each model type is used in order to describe one aspect of a
system. For example, system is described in terms of different views in case of visual
modeling language UML. Each view has one or more corresponding diagram types.
UML version 1.5 specifies nine types of diagrams [3] and UML 2.0 specifies thirteen
types of diagrams [4]. In the different projects only subsets of these diagram types are
used, depending on the goals. But more than one type of diagrams is needed in order to
describe the static structure as well as the behaviour of the system. For example,
Larman [5] presents a possible structure of a system analysis specification. The
specification must contain UML diagrams (visual models) as well as textual models:
• Use case model (diagrams and textual specification of use-cases).
• Domain model (diagrams and textual specification of conceptual classes and
attributes).
• System sequence diagram.
• Contracts of the system operations.
These models together also constitute a model. Final versions of the models that
describe a system have to be syntactically and semantically correct, complete and
consistent within itself and with each other in order to be most useful. If model is the
combination of diagrams and textual description, then there can be inconsistencies
between these components.
Examples of the inconsistencies within one model:
• Use cases/actors have different names in a diagram and in a text.
• There are different amount of use cases/actors in a diagram and in a text.
• A use case is associated with the different actors in a diagram and in a text.
Examples of the inconsistencies across different models:
• Names of the actors are different in a use case model and in the sequence
diagrams that describe system operations.
• Names of the elements of a domain model differ from the names that are used
in the pre- or post conditions of the contracts of the system operations.
Examples of the completeness problems:
• A domain model is missing.
• A use case diagram is not accompanied with a textual specification.
• An operation contract doesn't have the post-conditions.
Checking of the models in order to find such problems can be at least partially
automated by the software system. Models could be inconsistent and incomplete during
the development process. But it should be possible at any time to get information about
these problems. It helps to gradually improve the quality of the models. Pedagogical
pattern Built in Failure [6] suggests that teacher should remove the fear of failure as a
barrier to learning by making failure a part of the learning process. A hypothesis that
must be controlled in the future is that the checking functionality would also change a
modeling tool to a valuable learning tool. It is because continuous feedback from the
system instead of a teacher reduces fear to make mistakes that hampers the learning
process. It also eases the work of a teacher who doesn't have to check consistency and
completeness (CC) problems manually.
But CASE systems today do not provide enough support for checking consistency
between different types of models [7], [8] and between evolving versions of models
[9]. Instead it is mostly the manual work and takes quite a long time and a lot of effort.
And even if CASE tools provide some support it is still a "daunting tasks beyond
anyone’s cognitive ability" [10] because of big amount of different types of models.
An author of this article teaches database design in a university. A part of the
course work is a term project. Students have to create a strategic- and detailed analysis
of an information system and a prototype of its software. Current structure of the
project documentation has been used during the last four years. For example, 75
projects where presented in the spring of 2005. Teacher reviewed projects together with
their authors and pointed to the mistakes. Average length of a review of one project
was about 20 minutes. Average interval between first checking of a project by the
teacher and acceptance of the project was 4.2 days. Students improved their projects
during this period and sometimes they did it repeatedly. Students used word processor
in order to write textual models and CASE-tools or diagram editors in order to draw
visual models. These systems don't support automatic consistency and completeness
checks. Presented projects reflected this situation and contained many such
deficiencies. Another problem is that students have difficulties to understand how all
these different models are connected with each other. Therefore the system is needed
that gives fast and precise feedback to the students.
Delen et al. [8] write: "Second, there is a need for a completely Web-based
integrated modeling environment for distributed collaborative users." Our main
contribution is description of the system that allows to perform strategic- and detailed
analysis of a database-centric information system without using complicated visual
notations. This system should ease creation of an independent work by the students and
correction of it by the teachers. It can also be used in the real-world information system
development projects. System should support methodological framework for the
Enterprise Information System (EIS) strategic analysis (for example, framework that is
described by Roost et al. [1]). Person with the modeler role will record system
specification in a database as an integrated model using one tool. Modeler doesn't have
to create a collection of weakly connected models by using different tools any more.
Modeler will use the modeling language that is simplified synthesis of the different
system specification languages (UML [3], SMX [11] and OPM [10] among others).
The structure of the database will be derived from the fixed metamodel of this
language. We plan to use DBMS that allows to use SQL language. Because of the fixed
metamodel this system will be more similar to a CASE than to a Meta-CASE tool. Data
about the various aspects of a system will be recorded in a database using a form-based
web interface. We propose queries that find consistency or completeness (CC)
problems that are present in a specification. Our approach eliminates problems with the
inconsistencies between diagrammatic and textual representations. They are kind of
views to the information in the database that could be generated by the system at any
time. We don't have yet completely implemented the system but have started to create
its prototype. We see that such a system will help students/teachers in the
learning/teaching process and therefore continue its development. It will also help to
improve quality of the result of the real information system development projects.
The rest of the paper is organized as follows. Section 1 gives an overview of the
existing work about checking consistency and completeness of the models. Section 2
presents the metamodel of the modeling language that our system will use. Section 3
lists queries that help to identify consistency and completeness problems of the model.
Section 4 contains discussion and comparison of our work with the existing work in the
field. Section 5 summarizes this article.
1. Related Works
UML is nowadays de facto standard for describing the information systems. Problems
of consistency between UML diagrams have for example acknowledged by Engels and
Groewegen [12] who describe open issues in the object-oriented development.
Modeler has to learn a lot of different notations, rules and guidelines that are used
in the different types of models. McLeod [13] notes that UML 1.1 contains 233 discrete
concepts. Still he proposes rich visual notation for process models which could replace
UML dynamic diagrams [13]. If we study one UML or other similar model, then we
have to constantly look the other models in order to fully understand it. Switching
between different pages/ files/ packages is inconvenient as well as mentally challenging
and wearying. "Multiplicity of representational styles impedes communication between
modeling professionals and their clients." [14] Visual models are usually created with
the different CASE or model drawing tools (Rational Rose, ERWin, ArgoUML, Visio,
Dia etc.) and are accompanied with the textual specifications that are created using a
text editor. We have to have this software in our computer if we want to thoroughly
study these models or modify them. Modification of one model requires modifications
of dependent models as well. Multiplicity of modeling software and files that contain
models causes often creation of the model from scrach instead of reusing the existing
ones. Some approaches that are used in order to achieve correct and consistent models:
1. Formulation of the guiding rules that a modeler who creates different types of
models should follow. For example, Glinz [15] describes rules that help to
minimize inconsistencies between a class model and a use case model.
2. Usage of the cross references between the different types of models. For
example, Glinz [15] proposes to use references to a class model in the
scenarios of use cases.
3. Usage of the specific models which contain cross references between other
models. Example of such a model is a CRUD matrix. It shows associations
between object types and processes [16] and helps to check their consistency.
4. Usage of the systems that evaluate models created by the CASE tools or assist
user of a CASE tool.
5. Usage of the modeling notations and systems that use one type of model in
order to specify multiple aspects of a system.
Drawback of the approaches 2 and 3 is that without a tool support, references in a
model or new kinds of models may themselves have CC problems. CASE systems can
have supporting tools that check the models or provide active assistance to its users
(approach 4). They may transform the diagrams into some other form of representation
in order to analyze them. For example, generated description logics statements [9] are
analysed by using description logic query tool. Richters and Gogolla [7] describe the
system which translates UML models to the statements of UML-based Specification
Environment language. Models are then analyzed by simulating the case when model
elements have instances. Agents based system WayPointer [17] monitors the use case
model, which is created by some CASE tool, for completeness, consistency and
correctness. It can point to the problems and offer recommendations.
Agarwal and Sinha [18] conclude that developers don't rate any of the UML
diagrams as very high in terms of usability. It might be caused by the usage of the
several model types which leads to the inconsistencies between various parts of system
specification [10], [19]. Acknowledgement of the "model multiplicity problem" is not
new. A single model-based approach is superior to the multiple model approaches for
late requirements engineering through implementation according to Paige and Ostroff
[20]. Already Jäderlund [11] describes a methodology for the holistic system
development that uses so-called system matrices in order to describe the system.
System matrix (SMX) incorporates multiple views of the system. It provides methods
for checking correctness, completeness, and consistency (CCC check) of the system.
More recently model multiplicity problem has been addressed by introducing
Object-Process Methodology (OPM) [10], [19], which is a holistic system modeling,
development and evolution approach. OPM uses Object-Process Diagrams (OP
diagrams) for the graphic specification and Object-Process Language (OPL) for the
textual specification of the system. OPM uses one integrated type of model in order to
describe structural, functional and behavioural aspects of the system [10]. CASE tool
Object-Process Case Tool (OPCAT) that supports OPM has been developed [19].
Yet another example of the modeling languages that corresponds to the single-
model principle is Eiffel [20]. Delen et al. [8] propose system Modelmosaic that allows
to create different types of models. It records models and relationships between the
elements of different types of models in a single integrated information base. These
relationships, that are recorded as business rues, allow to generate new models from the
existing ones.
2. Description of the Modeling Language
A model is created by using some language. Specification of the semi-formal language
should contain descriptions of the abstract syntax, well-formedness rules and semantics
[21]. For example, metamodel that describes the abstract syntax of UML is presented
as a set of class diagrams [3], [4]. Well-formedness rules of UML are expressed using
OCL constraints and its semantics are described using free-form text. In this section we
present the metamodel of the language that will be used for specifying information
systems in our proposed system. Diagrams that present fragments of the metamodel are
accompanied with the free-form textual descriptions that explain some of the
underlying concepts. Structure of the database for recording specifications of the
information system will be derived from this metamodel. In the section 3 we present
queries that help to check the well-formedness of the recorded models.
Interested parties can participate in the information system development projects in
the different roles (see Figure 1).
An information system (IS) is described using three types of subsystems according
to the methodological framework for the Enterprise Information System (EIS) strategic
analysis [1]. These types are: areas of competence, functional subsystems and data
centric subsystems that are also called registers (see Figure 2). A functional subsystem
corresponds to one or more business processes [1]. "A register is a logical data-centric
view of a business object that holds the state and transactions data of the object and
provides related recording and query services." [1] Administrative subsystems help to
perform administrative tasks of the organizations. Examples are subsystems for the
management of data about the workers and documents. These kinds of subsystems are
part of many different information systems. Business subsystems help to perform
specific business tasks of the organizations. These tasks are reason why this
organization is founded in the first place. For example, university IS has subsystems for
the management of data about the students, curriculums and study results.
Functional subsystems use the services of one or more registers by reading and
modifying data in them. Subjects who have some role in an IS use the services of one
or more functional subsystems that belong to the area of competence of their role [1].
-name
Project
-first_name
-last_name
Person
Participant_role_type
Partipation_in_project
1 0..*
-participant
1
0..*
1
0..*
1..*
0..1
Information system
-name
Classifier
Party
-business_name
Organization
0..* 0..*
works for
-username
-password
User
1
0..1
Figure 1. Metamodel of the projects and participants
We provide possibility to specify non-functional requirements of the system (see
Figure 2) by using structure described in Volere Requirements Model [22].
Functional requirements of the information system can be specified as use cases.
Corresponding fragment of the metamodel (see Figure 4) is created based on the
guidelines of Larman [5] and Cockburn [23]. Each use case belongs to some functional
subsystem (see Figure 4).
Subsystem
Functional subsystem RegisterArea of competence
Task category
1
0..*
Information system
10..*
-user
0..* 0..*
For the administrative tasks of the organization
For the business tasks of the organization
1
0..*
Usage of register
0..* 0..*
1
0..*
Type of operation
Type of non-functonal requirement
-description
-rationale
-fit criterion
-customer satisfaction
-customer dissatisfaction
-supporting materials
Requirement
0..*
0..*
-text
System goal
1
0..*
-name
-background
System
Create
Read
Update
Delete
-name
Classifier
-name
Classifier
Look and Feel
Usability
Performance
Operational
Maintainability and Portability
Security
Cultural and Political
Legal
0..*
0..*
conflicting
-dependent0..*
0..*
Document
-contains related inf.
0..*
0..*
{Mandatory,
Or}
Figure 2. Metamodel of the subsystems
-name
Data element
-definition
Object type
Relationship
-Object type A
1
0..*
-Object type B
1
0..*
Register
1
0..*
Substance_type
0..1
0..*
Abstract object
Physical object
-name
Classifier
Multiplicity
-A1
0..*
-B
1
0..*
-a_is_generalization
-b_is_generalization
Generalization
-role_of_object_A
-role_of_object_B
-text_a_b
-text_b_a
-a_is_aggregate
-b_is_aggregate
-a_is_composite
-b_is_composite
Association
{Mandatory,
Or}
Figure 3. Metamodel of the data elements
Actor
-name
-goal
-short description
-performance suggestion
-is_essential_UC
Use Case
-description
Interest
1
0..*
1
0..*
-description
-problem identification time
Open_issue
10..*
-description
Action
0..*
1
0..*
-Performer
0..1
-extension
0..*
0..1
Functional subsystem
Area of competence
0..1
1
AgentInstrument
Actor_role_type
1
0..*
Primary actor
Supporting actor
Offstage actor
0..* 0..*
extension
-name
Classifier
0..* 0..*
inclusion
Action_at_any_time
-seq_nr
Action_in_sequence
included
extending
-description
-solving time
Solution
1
0..1
-text
-is_triggered_by_time
Event
1
0..*
-trigger
1
0..*
Scope
10..*
Inside the scope of IS
Adjacent (external)
ext. point
Activity type
1
0..*
-name
Classifier
Manual
Computer suported
Fully automated
Embedded
-value
UC metrics
1
0..*
UC metric type
priority ranking
max freq. during a day
avg freq. during a day
min. duration
avf. duration
max. duration
1
0..*
Figure 4. Metamodel of the use cases
-name
Data element
-name
Database operation
Type of operation
-name
Classifier
Postcondition
10..*
1
0..*
0..*
1
-description
Action
-result
0..*
-cause
0..*
Create
Read
Update
Delete
-value is input
0..* 0..*
-value
Operation metrics
1
0..*
Operation metrics type
Amt. per operation
Avg. amt. per hour
Max. amt. per hour
1
0..*
Figure 5. Metamodel of the database operations
-text
-is_triggered_by_time
Event
-definition
Object type
-name
-is_start_state
-is_final_state
State
State change
-trigger1
0..*
-before 1
0..*
-after1
0..*
1
0..*
{States before and after
are the states of the same
object type}
Figure 6. Metamodel of the state changes
Each use case describes scenarios that consist of actions. Most of actions are
performed sequentially. Some actions can be performed at any point of the scenario.
Use cases can be related by using either extension or inclusion relationships.
Actions may be performed by the actors who are either agents or instruments.
Agent corresponds to the area of competence. Registers that are used by a functional
subsystem are specified in terms of the data elements (see Figure 3) and contracts of
the database operations (see Figure 5). Action that is part of the use case can cause
execution of the database operation. Result of the operation is described in terms of the
post-conditions. Use case is triggered by an event (see Figure 4). Events cause change
of the state of the objects (see Figure 6).
3. Consistency and Completeness (CC) Checks
Structure of the database determines the elements that can be associated. It also
enforces the constraints of the metamodel according to which participation and/or
cardinality is 1 at some relationship end. General principle of our system is that
consistency and completeness of the models will not be ensured by the database
constraints. It gives more freedom to the modeler. Occurrences of each problem will be
found by using a query or a set of queries. Next we present some CC checks that the
system must be able to perform. A query must find the cases where the CC rule is not
fulfilled. Another possibility is to create a query that simply returns the value false if a
model contains a problem. It must be possible to use these queries at any moment.
1. An IS consists of at least one area of competence (AC), functional subsystem (FS)
and registry subsystems (RS).
2. An AC has exactly one corresponding actor.
3. An actor (and therefore an AC) uses services of at least one FS.
4. Services of a FS are used by at least one actor.
5. An IS (through some of its FS) is used by at least one non-adjacent actor.
6. A FS uses services of at least one RS.
7. A RS has at least one FS that reads its data.
8. A RS has at least one FS that adds new data to it.
9. An IS has at least one administrative FS and RS and business FS and RS.
10. An IS (as a whole or through some of its FS) has at least one non-functional
requirement from each of the different requirements types.
11. Functional requirements to a FS are described by using at least one use case.
12. A use case describes process what has between n and m actions (steps). Numbers n
and m should be dynamically changeable.
13. A use case is associated with the description about the interest of a primary actor
in the context of the use case.
14. A use case has ideally zero unsolved open issues.
15. A use case that is not an essential use case is associated with at least one database
operation through some action.
16. A data element is created and read by at least one database operation (association
through a post-condition).
17. A database operation has at least one post-condition.
18. A database operation is associated with at least one action that is part of the non-
essential use case. Each database operation is used by at least one use case.
19. Associations between FSs and RSs are consistent with the association of the use
cases (that belong to some FS) and data objects (that belong to some RS).
20. It is possible to get from the start state of an object to any other state of an object
by using state transitions.
21. An object type has exactly one start state and one or more end-states.
22. Relationships have mark about aggregation/composition/generalization at most at
one end. In this case other end has no such marks.
Queries can also be used in order to find suspicious parts of the specification that
may or may not be the mistakes. For example, query can search registers and data
elements that are not subject of the update or delete operations. Queries help to collect
metrics about the IS (like proposed by [24]). For example, we can count amount of use
cases in different FS-s. If some FS has a big amount of use cases compared to others
then it shows that the FSs are unbalanced and some big FS needs to be further divided.
4. Discussion and Comparisons
Commercial tools Modelmosaic [8] and EA WebModeler record models in a database.
EA WebModeler [25] provides form-based web interface for creating models. Our
system is not unique in this regard. UML is widely used notation and therefore it is
reasonable to teach it even after we start to use this system. For the teaching and
presentation purposes it is sometimes useful to see system specifications in the form of
UML diagrams. Possible solution is the program feature which generates XMI files that
contain models, based on the data in the database. These files can be opened using a
CASE tool. Similar feature exists in EA WebModeler [25].
A modelling language that follows the principle of the single model must satisfy
following three criteria: conceptual integrity, consistency of views, wide spectrum
applicability [20]. Our proposed solution satisfies "conceptual integrity" criteria
because models are recorded in one logical database. Each model element is recorded
only once. Our proposed solution satisfies "consistency of views" criteria because
checking of the consistency of different views of a model is automated. Our proposed
solution partly satisfies "wide-spectrum applicability" criteria. This system is used in
order to perform strategic and detailed analysis of the system but not design or
implementation. In the next iterations system could be extended in order to allow
generation of the stored procedures based on the database operations and table
specifications based on the data-elements.
Why can't we use existing free software in order to model systems by using single
model type? Examples of such systems are SystemSpecifier [26] that allows to create
system matrices and OPCAT [19] that allows to create OPM models. First reason is
that they don't fully support the methodological framework for the Enterprise
Information System (EIS) strategic analysis. They don't allow to specify different types
of subsystems and their interconnections. In addition, proposed system is different from
OPCAT and SystemSpecifier because models are recorded in a database but not
directly in the files. It helps to avoid well known problems of the file based systems
like separation-, isolation- and duplication of data. More than one modeler can work
with the same model at a time. Problem of OPCAT is that it doesn't provide explicitly
CCC check functionality. Our proposed system allows to use queries from the database
in order to find consistency and completeness problems. Usage of the integrated model
prevents repeating recording of the same information and thus helps to avoid
inconsistencies. Queries can be used in order to classify elements of a model. For
example, it is possible to determine whether a use case is concrete, abstract, base or
addition [5, p. 388] by making query about its relationships with the other use cases.
Another example is that we don't have to separately record the events that influence the
registers. Instead query can find the events that trigger the use cases which steps use
the services of the register. Our proposed system can also be used in order to collect
information about the work amount and performance of the modelers. It would also be
a useful e-learning tool, because it is planned to be a web-based system.
5. Conclusions
In this article we have described principles of the system that helps to perform strategic
and detailed analysis of the information systems. This system allows to record one
integrated model of the system into a database by using form-based web interface. It
provides queries for finding consistency and completeness (CC) problems of the
model. These queries can be used at any moment during the modeling process. We
have created a partial prototype of the system for managing subsystems (see Figure 2).
This prototype also allows to manage and execute queries (see Section 3).
If the system will be ready, then we can perform the usability study in order to
evaluate which way users prefer to describe the system – using visual diagrams with
little support to CC check or using textual descriptions with the extensive CC checks.
References
[1] M. Roost, R. Kuusik, K. Rava, T. Veskioja, Enterprise Information System Strategic Analysis and
Development: Forming Information System Development Space in an Enterprise. Proceedings of the
International Conference on Computational Intelligence, (2004) 215-219.
[2] J. A. Zachman, A framework for information systems architecture, IBM Systems Journal, Vol. 26, issue
3, (1987) 276 – 292.
[3] OMG Unified Modeling Language Specification. March 2003. Version 1.5. formal/03-03-01.
[4] OMG Unified Modeling Language Specification. August 2005. Version 2.0. formal/05-07-04.
[5] C. Larman, Applying UML and patterns: an introduction to object-oriented analysis and design and the
Unified Process. Prentice Hall, Upper Saddle River, USA, 2002.
[6] J. Eckstein, J. Bergin, K. Marquardt, M. L. Manns, H. Sharp, E. Wallingford, Patterns for Experimental
Learning, Proceedings of EuroPLoP 2001, (2001) Retrieved 16.03.2006 from
http://www.pedagogicalpatterns.org/current/experientiallearning.pdf
[7] M. Richters, M. Gogolla, Validating UML Models and OCL Constraints. UML 2000 (2000), 265-277.
[8] D. Delen, N. P. Dalal, P. C. Benjamin, Integrated modeling: the key to holistic understanding of the
enterprise. Commun. ACM 48, Vol. 4, (2005), 107-112.
[9] R. Van Der Straeten, T. Mens, J. Simmonds, V. Jonckers, Using Description Logic to Maintain
Consistency between UML Models. UML 2003 (2003) 326-340.
[10] D. Dori, Why Significant Change in UML is Unlikely, Communications of the ACM, Nov.2002, 82-85.
[11] C. Jäderlund, Systematrix. Complete SMX handbook. Stockholm, 1981.
[12] G. Engels and L. Groenewegen, Object-oriented modeling: A roadmap, Future Of Software
Engineering 2000, 105–116.
[13] G. McLeod, Beyond Use Cases, Proceedings of EMMSAD'00: 5th IFIP WG8.1 Int. Workshopon
Evaluation of Modeling Methods in System Analysis and Design, Kista, Sweden.
[14] A. M. Geoffrion, Computer-based modeling environments, European Journal on Operational Research
41 (1989), 33–43.
[15] M. Glinz, A Lightweight Approach to Consistency of Scenarios and Class Models. Proceedings of the
4th International Conference on Requirements Engineering (ICRE'00) (2000), 49-58.
[16] D. Brandon, Crud matrices for detailed object oriented design. J. Comput. Small Coll. 18, 2 (Dec.
2002), 306-322.
[17] R. Racko, A Cool Tool Tool, Software Development Magazine, May 2004, Vol. 12, Part 5, 21-26.
[18] R. Agarwal and A. P. Sinha, Object-oriented modeling with UML: a study of developers' perceptions,
Commun. ACM, Vol. 46, No. 9, (2003), 248-256.
[19] D. Dori, I. Reinhartz-Berger and A. Sturm, OPCAT - A Bimodal CASE Tool for Object- Process Based
System Development, Proceedings of International Conference on Enterprise Information Systems,
(2003), 286-291.
[20] R. Paige, J. Ostroff: The Single Model Principle. Fifth IEEE International Symposium on Requirements
Engineering (RE'01), (2001), 292-293.
[21] J. Greenfield, K. Short, Software Factories: Assembling Applications with Patterns, Models,
Frameworks, and Tools. John Wiley & Sons, USA, 2000.
[22] S. Robertson, J. Robertson, Mastering the requirements process. Addison-Wesley, USA, 1999.
[23] A. Cockburn, Basic Use Case Template, Versaion 2, October 26, 1998, Retrieved 11.03.2006 from
http://alistair.cockburn.us/usecases/uctempla.doc
[24] H. Kim, C. Boldyreff, Developing software metrics applicable to UML Models, Proceedings of the 6
th
International Workshop on Quantitative Approaches in Object–Oriented Software Engineering, (2002).
[25] Agilense Enterprise Architecture Frameworks. Retrieved 12.03.2006 from
http://www.agilense.com/documents/agilense_frameworks.doc
[26] Systematik holistik metodik. Retrieved 05.03.2006 from http://www.systematik.se/
