A MODEL DRIVEN ARCHITECTURE TOOL BASED

ON SEMANTIC ANALYSIS METHOD

Thiago Medeiros dos Santos, Rodrigo Bonacin

CESET, Unicamp and CenPRA, MCT, Rua Paschoal Marmo, 1888, Limeira, SP, Brazil

M. Cecilia C. Baranauskas

Institute of Computing, Unicamp, Caixa Postal 6176, 13083 970, Campinas, São Paulo, Brazil

Marcos Antônio Rodrigues

CenPRA, MCT, Rodovia Dom Pedro I, km 143,6, 13069-901, Campinas, São Paulo, Brazil

Keywords: Organizational Semiotics, Model Driven Architecture, Model Driven Engineering.

Abstract: Literature in Information Systems presents a set of approaches to analyzing and modeling the organizational

context in order to improve the quality of the computational systems. The research community also looks

for alternatives for aligning these approaches with Software Engineering techniques, which support the

large scale and low cost software production and maintenance. In this work is proposed an approach and a

tool to support the construction of computational models taken from the organizational models. This

approach is based on concepts, methods and techniques from MDA (Model Driven Architecture), and from

Organizational Semiotics, more specifically the MEASUR (Methods for Eliciting, Analyzing and

Specifying User’s Requirements). A MDA tool was constructed in order to realize the approach. The first

version of this tool provides means to model Ontology Charts. By using semi-automated transformations, it

is also possible to produce UML (Unified Modeling Language) class diagrams from the modeled Ontology

Charts. The tool features, design solutions, and capabilities are explained.

1 INTRODUCTION

The design and use of computational systems are

susceptible to the organizational context where they

are immersed. Literature presents a set of

approaches to analyze and to model the

organizational context in order to improve the

quality of computational systems.

The research community also looks for

alternatives for aligning these approaches with

Software Engineering techniques, which support the

large scale and low cost software production, and

maintenance. Among the main difficulties is the

multidisciplinary nature of the problem, which

require methods, techniques and tools coming from

many domains, including: humanities, to understand

and model the organizational context; engineering,

to deal with aspects related to large scale and low

cost production; and technologies, to construct

appropriate techniques and tools.

In this work we propose an approach and tool to

support the construction of computational models

derived from the organizational models. The main

objectives of the proposed approach are to allow the

representation of the organizational context

semantics, and to provide a way to quickly produce

corresponding design models.

In order to achieve these objectives, we make

use of concepts, tools, and methods from Model

Driven Architecture (MDA) (OMG, 2003) and

Organizational Semiotics (OS) (Liu, 2000). OS

research explores the use of signs and their effects

on social practices. OS understands that each

organized behavior is affected by the

communication and interpretation of signs by

people, individually or in groups. This work is based

on Stamper’s school of OS (Stamper et al., 1988;

Stamper 2001), more specifically the use of the

MEASUR (Methods for Eliciting, Analyzing and

Specifying User’s Requirements) (Stamper et al.,

305

Medeiros dos Santos T., Bonacin R., Cecilia C. Baranauskas M. and Antônio Rodrigues M. (2008).

A MODEL DRIVEN ARCHITECTURE TOOL BASED ON SEMANTIC ANALYSIS METHOD.

In Proceedings of the Tenth International Conference on Enterprise Information Systems - ISAS, pages 305-310

DOI: 10.5220/0001703703050310

 SciTePress

1988) to model the semantic aspects of the

organization.

The MDA is a proposal that permits multiples

refinement of abstract models to generate the

systems code, through transformations (Gardner et

al., 2003). This work explores how Computation

Independent Models (CIM) can be refined using the

MDA approach.

A MDA tool was constructed in order to

materialize the approach. The first version of this

tool supports the modeling of Ontology Charts

produced during the Semantic Analysis Method

(from MEASUR). By using semi-automated

transformations, it is also possible to produce UML

(Unified Modeling Language) class diagrams from

the modeled Ontology Charts. In this paper, the tool

features are explained trough a small example of

ontology chart.

The paper is organised as follows: Section 2

presents the Model Driven Architecture and the

Semantic Analysis Method, Section 3 discusses the

proposed approach to produce Platform Independent

Models from Computation Independent Models,

Section 4 presents Sonar, the developed tool, Section

5 discusses the approach and further works, and

Session 6 concludes.

2 BACKGROUND

On this section we present concepts and methods

from Model Driven Architecture (MDA) and the

Semantic Analysis Method (SAM).

2.1 Model Driven Architecture

MDA is an Object Management Group (OMG)

proposal to promote an open solution for the

challenges related to constant business and

technological changes. To reach this objective in

MDA the application logic and the technological

issues of a specific platform are specified in models

of different abstraction levels.

In order to introduce MDA, firstly is presented

what the OMG means by the following concepts:

– Architecture: The OMG adopts the Shaw and

Garlan (1996) concept. “The architecture of a

system is a specification of the parts and

connectors of the system and the rules for the

interactions of the parts using the connectors.”

(OMG, 2003, p.2-3) The MDA specifies kinds

of models and their relations.

– Platform: “A platform is a set of subsystems and

technologies that provide a coherent set of

functionality through interfaces and specified

usage patterns, which any application supported

by that platform can use without concern for the

details of how the functionality provided by the

platform is implemented.” (OMG, 2003, p.2-3)

– Viewpoint: “A viewpoint on a system is a

technique for abstraction using a selected set of

architectural concepts and structuring rules, in

order to focus on particular concerns within that

system.” (OMG, 2003, p.2-3)

– Model: “A model of a system is a description or

specification of that system and its environment

for some certain purpose.” (OMG, 2003, p.2-2)

In the MDA specification three types of

viewpoints are proposed, as well as one model for

each viewpoint: the CIM (Computation Independent

Model), the PIM (Platform Independent Model) and

the PSM (Platform Specific Model). The CIM,

which is the focus of this work, has an important

role as a bridge between developers and domain

specialists. The CIM specifications are requirements

for PIM and PSM.

These models are specified according to

metamodels (models that describe models) that

should be specified according to metametamodels.

The metametamodels should be self-described.

Other metamodel layers could also be created if

necessary. According to MetaObjectFacility 2.0

(MOF) specification (OMG, 2006), the fundamental

concepts can be used to handle any number of

layers.

In MDA one model can be converted to another

model of the same system (usually more specific)

through a transformation process. The

transformation follows mappings, which can be

understood as a set of rules and methods to obtain a

new model from a previous one. The Model Type

Mappings allows specifying transformations from

any model built using types specified in PIM to

models expressed using PSM types. The Model

Instance Mappings allows specifying model

elements which should be transformed in a particular

way.

According to Brown and Conallen (2005), some

times it is not possible to obtain automatic

transformations from model to model. The

transformation rules must be clear and unambiguous,

and should be able to programmatically access the

necessary elements of the models. Brown and

Conallen (2005) also emphasize that most of the

steps that involve natural language documents

reading are not typically suitable to MDA-style of

automation.

ICEIS 2008 - International Conference on Enterprise Information Systems

306

This is the main reason why automated

transformations from CIM to PIM are not common.

Usually, CIM models are based on textual

descriptions (i.e. Use Cases). However, better

structured CIM models, based on a visual language,

can help to construct systems aligned to business

and human factors, and even the “semi-automated”

transformations could bring benefits associated to

the cost, time and quality of the software. The next

section presents the Semantic Analysis Method as an

example of a CIM model constructed according to

Organizational Semiotics concepts.

2.2 Semantic Analysis Method

The Semantic Analysis Method (SAM) assists the

users or problem owners in eliciting and

representing their requirements in a formal and

precise model. With the analyst in the role of

facilitator, the required system functions are

specified in an ontology chart, which describes a

view of responsible agents in the focal business

domain and their behavior or action patterns named

affordances (Liu, 2000). Some basic concepts of

SAM adopted in this paper are based in Liu (2000):

– “The world” is socially constructed by the

actions of agents, on the basis of what is offered

by the physical world itself;

– “Affordance”, the concept introduced by Gibson

(1977), is used to express invariant repertories of

behavior of an organism made available by some

combined structure of the organism and its

environment. In SAM (Stamper, 1993) the

concept introduced by Gibson was extended by

Stamper to include invariants of behavior in the

social world;

– “Agent” can be defined as something that has

responsible behavior. An agent can be an

individual person, a cultural group, a language

community, a society, etc. (an employee, a

department, an organization, etc.);

– “An ontological dependency” is formed when an

affordance is possible only if certain other

affordances are available. The affordance “A” is

ontological dependent on the affordance “B”

means that “A” is only possible when “B” is also

possible;

– “Determiners” are properties which are

invariants of quality and quantity that

differentiate one instance from another;

– “Specialization”, agents and affordances can be

placed in generic-specific structures according to

whether or not they possess shared or different

properties;

The concepts of Semantic Analysis are

represented by means of ontology charts, which

have a graphical notation to represent agents

(circles), affordances (rectangles), ontological

dependencies (lines drawn from left to right), role-

names (parentheses) and whole-part relations (dot).

3 A PROPOSAL FOR

BUILDING CIM TO PIM

TRANSFORMATIONS

Ontology Chart can be understood as a CIM model

given that it focuses on organization and

requirements. However, the SAM addresses issues

that are not represented in any of the UML diagrams

and it provides a different way of thinking about the

organization if compared with the Object Oriented

paradigm. Consequently the transformation from

Ontology Chart to Class diagrams is not a trivial

task. It is also possible to see the Ontology Chart as

a kind of CIM metamodel (not adopted in this

work), which specifies the semantics of other

models.

The rationale to construct the transformation

rules is based on a previous work that deals with

how to build UML Class Diagram informed by

SAM (Bonacin et al., 2004). Nevertheless,

additional aspects must be considered in CIM to

PIM transformations. The following steps are

proposed to build CIM to PIM transformations: (1)

create the transformation rules, (2) develop the

metamodels for CIM, (3) analyze which rules need

human intervention, (4) create the heuristic

metamodels, (5) design the interfaces to facilitate the

human intervention, and (6) develop the

transformations.

Regarding the rule creation, in this work was

adopted a group of ten heuristic rules proposed in

Bonacin et al. (2004). The next step was the

development of meta-models for ontology charts.

The KM3 (Kernel MetaMetaModel) language

(Atlas, 2005) facilitated the description of

Metamodels in MOF (MetaObject Facility) (OMG,

2006), and the Ecore (Budinsky et al.; 2003).

Figure 1 presents an UML diagram for the

Ontology Chart Metamodel. The concept of

“element” was created to describe anything at the

diagram. Affordance is a central concept at the

diagram. Role-name and agents are special types of

affordances. Affordances may also include

determiners and specializations. Ontological

dependencies link affordances.

A MODEL DRIVEN ARCHITECTURE TOOL BASED ON SEMANTIC ANALYSIS METHOD

307

Figure 1: Ontology Chart Metamodel.

Whole-part is a special type of ontological

dependency and an ontological dependency may

also include a role-name.

After constructing the Metamodel, the proposal

was to find out the heuristics that could be directly

automated, and which rules need human

intervention. An example of a non-directly

automated transformation is the creation of classes

or operations from affordances names. It is

necessary a human intervention to say whether an

automated rule should be applied or not, and to

decide which automated rule should be applied: to

transform the affordance into a class or operation.

If the second option is selected, additional

information is necessary to construct the PIM.

Figure 2 shows an intermediary model proposed to

describe this information. It has the “HAffordance”

Class with three attributes: Operation indicates if the

affordance will be mapped to an operation or to a

class, ClassId indicates the class that will contain the

operation, and affordanceId identify the affordance

that will be mapped.

The next step is the design of the interfaces for

end-user intervention. A wizard based solution was

adopted in this work. The last step is the

implementation of the transformations. They map

data from the ontology chart metamodel and the

heuristic model to PIM (UML models).

4 SONAR TOOL

As Figure 3 shows, the Ontology Chart modeling

tool, named Sonar, has the basic features of a CASE

(Computer-Aided Software Engineering) tool,

including: icons to model the diagram elements,

copy, paste, undo, redo, zoom, and export as image.

The fragment of the ontology chart diagram in

Figure 3 will be used to explain how the

transformation occurs internally in this modeling

tool. This fragment has the “society” as the root

agent; “person” and “thing” is ontologically

dependent of “society”; and “owns” is ontologically

dependent of “person” and “thing”.

Figure 2: Example of Heuristic Metamodel.

Figure 3: Sonar Main Window.

After modeling the Ontology Chart, the user can

transform it to UML models by using wizards on the

“Tool” menu. Figure 4 shows an example of wizard

where designers can specify which affordance will

be mapped to operation, and associate them to

classes (affordances mapped to classes) that will

contain it. For example the class “Person” contains

the method “owns”. By associating “owns” to

“person” a heuristic model is created following the

heuristic metamodel specification. The heuristic

ICEIS 2008 - International Conference on Enterprise Information Systems

308

model contains the necessary data to determine the

appropriated transformations that will be applied.

Figure 4: Example of wizard to human intervention.

Figure 5 shows an overview of the

transformations implemented by the Sonar tool. At

the CIM level there are two models: the ontology

model constructed according to the diagram, and the

heuristic model that is a result of human intervention

through wizards. These models follow the

metamodels specified in MOF and/or Ecore.

Internally the models are represented according to

the XMI format (OMG, 2007).

Transformations (Figure 5) were constructed

using the ATL (Atlas Transformation Language)

(Atlas, 2006). “ATL provides developers with a way

to produce a number of target models from a set of

source models.” (Atlas, 2006, p.1).

Figure 5: Sonar Transformations.

There are two source models (ontology and heu-

ristic) that produce target models. After producing

the PIM a number of transformations, can be applied

to produce PSM models, such as: to C++, to EJB,

and to SQL. According to Figure 5, in this version of

Sonar we can transform the UML to a Java model

and finally to Java code.

Figure 6 is a graphical representation of a

possible UML model produced from the ontology

model presented at Figure 3. This model contains

three classes produced from the “Person”, “Thing”

and “Society”. The user has decided to transform the

“Owns” affordance in a method of the class

“Person”. The method could return, for example, the

thing(s) that the “Person” owns. Alternative models

can be produced according to the human

intervention, such as, one model without the

“Society” class.

Figure 6: Constructed UML Model.

5 DISCUSSION AND FURTHER

WORK

The first version of Sonar can be understood as a

starting point to further research. One important

feature of the proposed approach is the possibility of

following the process: the capacity to detect the

origin and follow the tracking from the source model

to the produced model. Although the process

includes human intervention, the decisions are

registered on the heuristic models. Thus, it is

possible to verify how elements of the PIMs were

constructed from the CIM elements.

However, Ontology Charts do not produce a

complete PIM; consequently the PSM is also

incomplete. In a real life situation, the produced PIM

should be complemented with many details before

the transformation to PSM. In addition, important

decisions need human intervention (i.e. if implicit

elements of the CIM, like “society” should or not be

explicit in the implemented system). Nevertheless, it

is necessary to consider two main aspects:

– The produced model can be a valuable starting

point for the design of a complete and consistent

PIM. It produces a set of domain classes that are

closely related to the real context. Usually design

models closely related to real contexts result on

systems more reusable and easier to maintain;

– The approach can also be applied to equally well

define CIM (in addition to Ontology Chart), to

produce a more complete PIM, e.g.: a “norm

based model”, from the OS norms analysis

A MODEL DRIVEN ARCHITECTURE TOOL BASED ON SEMANTIC ANALYSIS METHOD

309

method (Stamper et al., 1988), could be used in

order to model the dynamic aspects of the

organization, consequently improving the

produced PIM.

It is also important to consider the possibility to

produce other diagrams from the Ontology Charts.

Ontology has been used to deal with problems of

systems integration. The Semantic Web (W3C,

2007) makes use of ontology to process the content

of information. The Ontology Charts could be used

to produce models based on technologies such as:

OWL (Web Ontology Language) (W3C, 2004). As

Semantic refers to meaning, and meaning are

socially created by humans, is expect to create a

more faithfulness Ontology if we start from a

Computer Independent Model.

6 CONCLUSIONS

The design and use of computational systems are

susceptible to the organizational context where they

are immersed. In this paper is proposed an approach

to produce computational models transformed from

the organizational ones. This approach has its basis

on techniques and methods from MDA and SO.

An Ontology Chart modeling tool, named Sonar,

implements the main proposed ideas. It shows the

possibility of generating PIM from CIM, using semi-

automated transformations. Among the main

benefits are the visibility of the transformation

process, and the production of an initial design

model closely related to the real context concepts.

ACKNOWLEDGEMENTS

We thank colleagues from CESET and IC at

Unicamp, and CenPRA for valuable contributions.

REFERENCES

Atlas group, 2005. Kernel MetaMetaModel v.0.3.

Retrieved December, 13, 2007, from http://www.eclip-

se.org/gmt/atl/doc/KernelMetaMetaModel%5Bv00.06

%5D.pdf

Atlas group, 2006. ATL User Manual.v. 0.7. Retrieved

December, 13, 2007, from http://www.eclipse.org/

gmt/atl/doc/ATL_User_Manual%5Bv0.7%5D.pdf

Bonacin, R., Baranauskas, M. C. C., and Liu, K., 2004.

From Ontology Charts to Class Diagrams: semantic

analysis aiding systems design. Proceedings of 6th

International Conference on Enterprise Information

Systems, Porto, Protugal.

Brown, A., and Conallen, J., 2005. An introduction to

Model Driven Architecture Part II: Lessons from the

design and use of an MDA toolkit. The Rational Edge.

Retrieved December, 13, 2007, from http://

www.ibm.com/developerworks/rational/library/apr05/

brown/index.html

Budinsky, F., Steinberg, D., Merks, Ed, Ellersick, R.,

Grose, T. J., 2003. Eclipse Modeling Framework: A

Developer's Guide, Addison Wesley Professional.

Gardner, T., Griffin, C., Koehler J., and Hause, R. A

review of OMG MOF 2.0 Query / Views /

Transformations Submissions and Recommendations

towards the final Standard. IBM Zurich Research

Laboratory, Retrieved December, 13, 2007, from

http://www.zurich.ibm.com/pdf/ebizz/gardner-etal.pdf

Gibson, J.J, 1977. The Theory of Affordances. In

Perceiving, Acting, and Knowing, Eds. Robert Shaw

and John Bransford.

Liu K. (2000). Semiotics in information systems

engineering, Cambridge. University Press.

Object Management Group (OMG), 2003. MDA Guide

Version 1.0.1. Eds. J. Miller and J. Mukerji. Retrieved

December, 13, 2007, from http://www.omg.org/

docs/omg/03-06-01.pdf

Object Management Group (OMG), 2006. Meta Object

Facility (MOF) Core Specification, Version 2.0.

Retrieved December, 13, 2007, from http://

www.omg.org/docs/formal/06-01-01.pdf

Object Management Group (OMG), 2007. MOF 2.0/XMI

Mapping, Version 2.1.1. Retrieved December, 13,

2007, from http://www.omg.org/docs/formal/07-12-01.pdf

Shaw, M. and Garlan, D., 1996. Software Architecture:

Perspectives on an Emerging Discipline. Prentice Hall.

Stamper, R. K., Althaus, K. e Backhouse, J., 1988.

MEASUR: Method for Eliciting, Analyzing and

Specifying User Requirements. In Olle, T. W.,

Verrijn-Stuart, A. A. e Bhabuts, L. (eds.)

Computerized Assistance During the Information

Systems Life Cycle. Elsevier Science, Amsterdam, 67-116.

Stamper, R., 1993. Social Norms in Requirements

Analysis – an outline of MEASUR. In Requirements

Engineering, Technical and Social Aspects, Eds. M.

Jirotka, J. Goguen, and M. Bickerton, Academic Press,

New York.

Stamper, R. K., 2001. Organisational Semiotics:

Informatics without the Computer? In Information,

Organisation and Technology: Studies in

Organisational Semiotics, eds. K. Liu, R. Clarke, P. B.

Andersen and R. K. Stamper. Kluwer Academic

Publishers.

W3C, 2004, OWL Web Ontology Language Reference,

Retrieved December, 13, 2007, from http://

www.w3.org/TR/owl-ref/

W3C, 2007, Semantic Web Activity, Retrieved December,

13, 2007, from http://www.w3.org/2001/sw/

ICEIS 2008 - International Conference on Enterprise Information Systems

310