2013_Non-diagrammatic_method_and_multi-representation_tool_Grigoriev_Kudryavtsev_CBI.pdf

8/14/2019 2013_Non-diagrammatic_method_and_multi-representation_tool_Grigoriev_Kudryavtsev_CBI.pdf

1/6

2013 IEEE International Conference on Business Informatics

258

Non-diagrammatic method and multi-representation tool for integrated

enterprise architecture and business process engineering

Lev GrigorievBusiness Engineering Group

Saint-Petersburg, Russia

[email protected]

Dmitry KudryavtsevSaint-Petersburg State Polytechnic University

Business Engineering Group

Saint-Petersburg, Russia

[email protected]

Abstract Nowadays enterprise architecting and business

process engineering are almost synonymous to diagramming.

Diagrams have many benefits and are sufficient for many

situations. But as the number of diagrams and their types

grows, they overlap and evolve, then it becomes hard to

maintain a collection of interrelated diagrams, even with the

help of a common repository. Besides the very nature of

enterprise architecting requires a lot of matrices (goals-processes, capabilities-processes, processes - applications).

The paper presents the multi-representation approach for

enterprise architecture model development and maintenance.

Classifications and matrices are the basis of this approach and

support knowledge structuring and integration. The ORG-

Master tool joins classifications and matrices with traditional

diagram-based technologies. The paper pays special attention

to the role of classifications and matrices in integrating

business processes into enterprise architecture

Keywords- enterprise architecture tool, business process

engineering, enterprise modeling tool, matrices, diagrams,

matrix method, multi-representation technology

I. INTRODUCTIONEnterprise architecting and business process

engineering are proven approaches for systemic and

sustainable organizational development. They are supported

by corresponding tools [1, 2, 3]. These tools document

business processes, enterprise strategies, business

capabilities, organizational structures, information

technologies and especially their interaction and

dependencies. Enterprise architecture management (EAM)

and business process analysis (BPA) tools not only capture

relevant information, but also process this information, e.g.

using reports, visualizations or applying analytical methods.

The model development interface is the most obvious part

of an EAM/BPA tool. It is the interface used to design,build, maintain and often manipulate the models that make

up the architecture.

Generally, models are built and maintained graphically

using node-link diagrams (a type of representation scheme,

that depicts elements or parameters of interest as nodes,

with arrows to indicate the connections). The tools model

development interface may also use textual interfaces to

allow additional information to be appended to the graphical

models. Diagrams have many benefits and are sufficient for

many situations in management [4, 5]. Diagrams work well

for enterprise architecture views, which express the EA

from the perspective of specific concerns and stakeholders

(in terms of [6]). They are also an essential result of any

EAM initiative and are provided to different stakeholders.

Unfortunately diagrams are not so good when there is

a need to develop and maintain a holistic and consistententerprise architecture model, which integrates a system of

partial views, especially under the following conditions:

evolving large-scale models, different and overlapping

diagram types, many diagram instances. Chief

architects/modelers, who support a whole system of

diagrams, may benefit from matrix-based representation.

There are two papers [7, 8] that compared the representation

power of matrix and node-link diagrams. Ghoniem et al. [7]

showed that matrices outperform node-link diagrams for

large or dense graphs in several low-level reading tasks,

except path finding. This difference is supported by user

study experiments conducted by Keller et al., as they found

that node-link diagrams offer better visual representation for

small, uncomplicated entities, but they are a complex formof representation for large systems and propagation across

systems [8]. Besides, two such researchers, Bidgood and

Jelley [9], gave compelling evidence that a matrix editor is

needed to combine processes with information architecture.

In their case study, they used a spreadsheet application as a

makeshift matrix editor, putting business processes on one

axis and business data on the other, and entering in the

relevant cells values indicating whether the process created,

read or updated the data.The paper presents multi-representation technology,

called ORG-Master, which was initially designed forbusiness architecture engineering and seamlessly integrates

complex business process models into enterprisearchitecture. Since classifications and matrices are the basisof this technology, the next chapter provides an overview ofmatrix-based methods. Chapter III explains the overallORM-Master modeling approach. Chapter IV describesmetamodeling language of the suggested tool. Chapter Vdescribes the tool, its matrix editing functionality and generalarchitecture (incl. integration with diagrammatic modeling).The role of matrices for business process integration into EAis described in chapter VI.


2/6


259

II. RELATED WORKA.History of matrix methodsKelly [10] provided a good overview of matrix methods.

The following chapter exposes Kellys main results:

Within the field of information systems development,

the use of the term 'matrix' is somewhat different from that

used in mathematics. Both share the concept of a grid ofvalues, whose location can be specified by their co-

ordinates, an enumeration along the horizontal and vertical

axes. In ISD, this is then extended by using objects, rather

than whole numbers, as each item on an axis. These objects

possess properties, and there may be some hierarchical

structure above the objects. Each element of the matrix can

be any (result of a function on some) item(s) present in the

model, and is determined by a function involving the items

that exist on the axes for its row and column.

Many early software design methods had or have

evolved matrix representations. These largely fell into

disuse, the graphical approach being preferred and

fashionable at that time. The development of graphical userinterfaces further favored the use of the diagrams in CASE

tools, and hence they mainly supported the graphical side of

these methods. The matrix side is still, however, as useful as

it was originally. Technology has now advanced, especially

in the area of databases, and can support a single conceptual

graph represented as both a diagram and a matrix, so the

user can benefit from the matrix representation in just those

areas where it is an easier paradigm than graphical

diagrams.

Further, in different fields various methods have been

developed that rely heavily on matrices: IBM's Business

Systems Planning (BSP) [11] for strategic information

system planning; Quality Function Deployment (QFD) [12]for product design and total quality management; Design

structure matrix (DSM) [13] for system decomposition and

integration problems in systems engineering of products,

processes, and organizations.

Support for these methods is typically limited to

proprietary, fixed-method tools, often produced by the

organisation that developed the method. This approach

reduces the flexibility of the program and the ability of the

user to do things his way to best suit the current

contingency.

The uses of matrices in methodologies can be divided

into three classes: those methods that use matrices as an

integral part, such as IBM's Business Systems Planning

[11], those that mainly use diagrams but where the use of

matrices as an alternate representation is recognized, such as

Two-Entity Data Flow Diagrams [14], and methods for

which there is no explicit use of matrices, but for which

applications of matrices can profitably be found.

Kelly [10] looks at these methods in more detail.

B.Matrix and table representations in modern EAM toolsand metamodeling platforms

Matrices are also actively used in business process

engineering and enterprise architecture management. For

example see the CRUD-matrix (Create, Read, Update, and

Delete), which links behavioral elements with information,

or TOGAF-recommended matrices: Actor/Role Matrix,

Application/Data Matrix, Application/Function Matrix etc.

Matrix representation is supported in the majority of

EAM tools:

Matrix Editor in ARIS Design Platform [15],

Matrix Editor in IBM System Architect [16],

Matrix Manager in Casewise Modeler [17].

Metamodeling platforms also support matrix and tabular

methods:

Matrix View and Tabular View in the ADOxx

Metamodelling Platform [18],

Matrix and Table Editors in MetaEdit+ Modeler

[19].

The main disadvantage of these matrix tools is that they

do not support grouping on the axis / hierarchical axis. Thesetools display only names of the objects instances of two

user-selected types in the repository. This is a crucial

limitation, since it doesnt allow working with large amount

of data; long lists cannot be processes by human brain. It is

especially harmful because matrices and spreadsheets are

often used to deal with long lists and Excel is one of the

main tools in many offices.

III. ORG-MASTER MULTI-REPRESENTATION MODELINGAPPROACH

The ORG-Master modeling approach has originally been

conceived in the course of the development of the business

engineering toolkit in 1998 [20]. Fig. 1

Non-diagrammatic method and tool are at the center ofthis technology. The modeling process starts fromknowledge acquisition. Enterprise modelers collect

represents thesuggested modeling technology.

Fig. 1: ORG-Master modeling approach


3/6


260

information about an enterprise from various sources(people's memory, documents etc.), then organize it usingdiagrams, classifications and matrices. Diagrams can becreated in ORG-Master graphical editors, which are based onMicrosoft Visio. These artifacts can be discussed and agreedupon with managers (managers can also make models bythemselves). This step is standard. Then ORG-Master

integrates all the acquired knowledge using classifications(hierarchical lists) and matrices so called internalrepresentation. These integration and complex structuring istypically done by highly qualified modelers; however theresultant integrated model is inappropriate for final users andenterprise stakeholders, so ORG-Master provides capabilitiesto specify and generate partial views from this model(diagrams, text and table reports), which will suit variousconcerns of various stakeholders.

IV. CLASSIFICATIONS AND MATRICES:THE ORG-MASTER METAMODELING LANGUAGE

Enterprise models are created using modeling languages,which are reflected in a metamodel. The creation of ametamodel is also done by using a modeling language. Thismodeling language is called the metamodeling language andit indirectly defines the way of modeling [21]. The basicbuilding blocks of ORG-Master metamodeling language(Fig. 2

Classification/hierarchical list - the representation

format for entities, hierarchical relationships between them

and values for the properties of entities. Hierarchy is the

main feature of classifications, and the main relationship

types are part-of, subclass of, be subordinated to etc.

) are:

Matrix - the representation format for relationships

between entities from classifiers. Example relationship types

are perform, help achieve, etc. The major part of

matrices links 2 classifiers, but n-ary matrices are also

supported. Inference matrix, allows to infer implicit

knowledge via the transitive property of relationships in

several matrices,

Types descriptions - specify the object types (classes),relationships and reusable attributes of metamodel, together

with their taxonomies.

Fig. 3 represents an example of classification: objects

(positions) are listed in the window on the left, and

pictograms define their types. The window on the right

exhibits the taxonomy of types, which can be assigned to

the objects in the current classification (see lower right

property window).

Fig. 4 represents an example of matrix, which is

visualized in the form of two lists. This matrix shows

relationships between classification (hierarchical list) of

Fig. 3. Classification for storing and structuring objects (e.g. Functional systems)

Fig. 2. Metamodeling language description


4/6


261

Fig. 4. Matrix allows to distribute Functional Processes between Functional systems

Functional systems and Functional processes, e.g. the

selected position -Administration and improvement of

marketing system consists of Process architecture

engineering..., Functional processes engineering andother functional processes.

V. THE ORG-MASTER MULTI-REPRESENTATION TOOLThe ORG-Master tool for business architecture engineering

provides strong matrix-editing capabilities. ORG-Master

includes the following modules: enterprise model editor,

reporting and query module, diagram editor and meta-editor

(Fig. 5

Classifications/matrices editing capabilities in the system

work together with diagramming (graphical editor, graphical

reports), which is a hybrid multi-representation technology.

The main module - enterprise model editor applies

classifications and matrices to knowledge acquisition (e.g.

via collaboration with managers), structuring andintegration. It produces large-scale enterprise models (e.g.

some example classifications in real-life models include

10000-20000 elements; corresponding matrices have even

more relationships). Reporting and query modules enable

specification of structure and format of external text and

table documents, then generate these documents. This

module also includes capabilities for diagram generation

based on model transformation (similar to [22]). These

capabilities support predefined notations and depend heavily

on the metamodel whether it is for commercial

organization, or for public administration, etc. Graphical

reporting includes automatic layout, but some complex

diagrams require additional manual improvements. The

recent project with Moscow city administration [23]includes 7 types of diagrams (mostly node-link); several

types of diagrams have about 1000 instances. Besides,

ORG-Master includes a MS Visio-based diagram editor,

which helps to acquire knowledge using diagrams,

transform it into classification/matrix form and back (works

with the common base) and enables to create complex

diagrams for final users based on the information from the

integrated model.

).

Fig. 5. The ORG-Master tool architecture


5/6


262

Metamodel development and customization can be done

using meta-editor you can define object types (classes),

relationship types and reusable attributes, then assign them

to classifications and matrices. This module is being

improved now in order to integrate principles of ontology

engineering [24, 25] and Domain-Specific Modeling [26,

27]. The current version of meta-editor partially supports

import from and export to Web Ontology Language (OWL).

Although metamodel is fully customizable, ORG-Master

methodology implies standard metamodels for certain types

of organizations (commercial company, public

administration, city). The next chapter

In order to distribute enterprise modeling artifacts portal

can be used, which provides integrated hypertext and

stakeholder-oriented representation of the enterprise model.

The portal is an addition to modeling tool.

exposes a fragment

of commercial company metamodel for illustrative purpose,

but a holistic description of metamodels is out of the scope

of the current paper.

The enterprise model editor, which supports

classifications and matrices, provides the followingfunctionalities in editing mode: group together related

entities in classifications; show/collapse subelements; sort

elements in classifications; 2 modes for matrix visual

representation (matrix as two lists, matrix as a grid),

show relationships for the selected element in a matrix

(matrix as two lists mode), support for directions of

relationships, assign attributes to objects in classifications

and to relationships in matrices, highlights, work with

groups of elements, etc. This functionality provides

analytical capabilities at the modeler's fingertips, for

example, to see objects without any relationship. Horizontal

layout of classifications in a matrix (matrix as two lists)

enables showing the hierarchical axis and long real lifenames of elements. The original, matrix as a grid, can

provide an adhoc overview of existing relationships. The

main differentiating feature of the suggested technology

from the other EAM matrix tools consists of the

combination of hierarchical lists with matrixes.

VI. ROLE OF CLASSIFICATIONS AND MATRICES ININTEGRATING BUSINESS PROCESSES INTO ENTERPRISE

ARCHITECTURE

The ORG-Master methodology uses classifications and

matrices to integrate processes into overall architecture. Fig.

6

represents a fragment of enterprise model, which exposes

process integration into business architecture (links with ITarchitecture are omitted in the example).

Fig. 6. Enterprise model as a system of classifications and matrixes

Enterprise functional decomposition [28] is modeled

using a chain of classifications and matrices. Functional

processes are structured in the corresponding classification

and compose functional systems through the matrix(see Fig. 4). Functional systems, in turn, are building blocks

for corporate and business systems. For example, some of

the functional systems compose end-to-end value-creating

processes, which are the main components of business

system. Matrixes foster reuse in functional decomposition,

e. g. one functional process can be reused in several

functional systems (see Fig. 4

Process goals and requirements can be linked to higher-

level goals and prioritized using QFD-like techniques

(Quality Function Deployment) [29; 30]. See Strategydeployment area in

); one functional system can

be reused in several end-to-end processes, etc. Besides

various classification principles can be used for various

activity elements, when we use several classifications.

Fig. 6

Matrices are also used to set responsibilities for the

activity elements (see matrices with Roles structure

in

. Classifications include goal

decomposition through aggregation relationship, while

matrices represent means-end relationships. Activity

elements are linked to goals through matrices.

Fig. 6

Information- and control-flow relationships can also be

stored and analysed (e.g. [31], which is not currently

supported by the ORG-Master) using non-diagrammatic

representation - see matrices with Results and

Documents in

). The RACI approach can be used here.

Fig. 6

The suggested method and tool have many large-scale

industrial implementations within business engineering

projects and public administration, which are described in

[20, 23].

(but such information is acquired and

distributed using standard node-link diagrams IDEF0,EPC, etc.).


6/6


263

VII. CONCLUSIONDiagramming is a predominant approach in business

process engineering and enterprise architecting today.

Although diagrams are very effective for knowledge

acquisition and communication, matrix-based methods

outperform them in some situations. Integration of large-

scale enterprise models and support for matrix methods(QFD, design structure matrix, etc.) are the examples. The

paper described method and tool (ORG-Master technology

in total), which unite matrix-based representations with

diagramming. ORG-Master uses classifications and matrices

for internal knowledge representation (for professional

modelers) and diagrams for external representation (for

communication with various stakeholders).

Advantages of the suggested matrix editing

technology: provides big picture (context) and details

simultaneously; combine model development with model

exploration and analysis (e.g. gap analysis: highlighted

elements without links in a matrix); compactness; large

amount of information can be clearly arranged and

manipulated at once. This allows the user to make a better

decomposition of the problem area into subsystems.

Application of the matrix-based technology was

demonstrated for business processes integration into

business architecture.

ACKNOWLEDGMENT

The authors would like to thank their colleagues Valentina Kislova, Tatiana Ouerdani, Alexey Zablotskiy,Evgenya Markelova and others for an enormouscontribution to the development of the ORG-Mastertechnology. Special thanks to Dmitry Koznov from SPbSUfor the useful feedback about the method and tool.

REFERENCES

[1] R. Scott Bittler, Magic Quadrant for Enterprise Architecture Tools,ID G00234030, Gartner Inc., 31 Oct. 2012, 28 p.

[2] S. Buckl, Ch. M. Schweda, On the State-of-the-Art in EnterpriseArchitecture Management Literature, June 1, 2011, 136 p.

[3] D. Norton, Magic Quadrant for Business Process AnalysisTools, IDG00219247,Gartner,Inc.,12 December 2011,22 p.

[4] R., Lengler, M. Eppler Towards a Periodic Table of VisualizationMethods for Management, Proc. of the Conference on Graphics andVisualization in Engineering, 2007, pp. 1-6.

[5] T. Gavrilova, D. Kudryavtsev, Diagrammatic knowledge modelingfor managers ontology-based approach, // Proc. InternationalConference on Knowledge engineering and Ontology Development,26-29 October, 2011, Paris, France. pp. 386-389.

[6] ISO/IEC 42010: 2011-Systems and software engineeringRecommended practice for architectural description of software-intensive systems. Technical report, ISO, 2011.

[7] M.Ghoniem, J. Fekete, P. Castagliola, On the readability of graphsusing node-link and matrixbased representations: a controlledexperiment and statistical analysis, Information Visualization, 4, no2, 2005, pp. 114135.

[8] R. Keller, C. M. Eckert, P. J. Clarkson, Matrices or node-linkdiagrams: which visual representation is better for visualisingconnectivity models?, Information Visualization, 5, 2006, pp. 6276.

[9] T. Bidgood, B. Jelley, Modelling Corporate Information Needs:fresh approaches to the information architecture, Journal of StrategicInformation Systems, vol. 1, No 1, Dec. 1991, pp. 3842.

[10] S.Kelly, A matrix editor for a metaCASE environment. Repr.from:Information and Software Technology,Vol.36, Is. 6, June 1994, pp,361371.

[11] IBM Corporation Business Systems Planning Information SystemsPlanning Guide, Publication #GE20-0527-4, 1975.

[12] Y. Akao, Quality function deployment: integrating customerrequirements into product design, Productivity Press, 2004.

[13] S. D. Eppinger, T. R. Browning, Design structure matrix methodsand applications. MIT Press, 2012.

[14] D. V. Steward, Software Engineering with Systems Analysis andDesign, Brooks/Cole, 1987.

[15] R.Davis, ARIS Design Platform Advanced Process Modelling andAdministration, London: Springer-Verlag London Limited, 2008,XVIII, 408 p.

[16] IBM Rational System Architect v. 11.4, Information center, viewed15 March2013, http://publib.boulder.ibm.com/infocenter/rsysarch/v11/topic/com.ibm.sa.help.doc/topics/c_Matrix_Editor.html

[17] Casewise Modeler, viewed 15 March2013, http://www.casewise.com/products/modeler

[18] H. Khn, "The ADOxx Metamodelling Platform." In Workshop onMethods as Plug-Ins for Meta-Modelling, Klagenfurt, Austria. 2010.

[19] MetaCase, Matrix Editor, viewed 15 March,2013, http://www.metacase.com/mep/matrix_editor.html

[20] L. Grigoriev, D. Kudryavtsev. The ontology-based businessarchitecture engineering framework, Proc. 10th InternationalConference on Intelligent Software Methodologies, Tools andTechniques (SOMET), September 28-30, 2011, Saint-Petersburg,Russia. 2011. pp. 233-252.

[21] D. Karagiannis, H. Khn, Metamodelling Platforms. Lecture Notes inComputer Science (2002), Springer-Verlag, p. 182.

[22] S. Buckl, A. M.Ernst, J. Lankes, C. M.Schweda, A. Wittenburg,"Generating visualizations of enterprise architectures using modeltransformations." EMISA 2007, 2007, pp. 33-46.

[23] A. Kostyrko, D. Kudryavtsev, L. Grigoriev, V. Kislova, A. Zhulin, L.Sinyatullina, R. Ermakov Regional enterprise architecture (EA)modeling for systemic development of city's ICT, Proc The 3rdRussian Conference on Knowledge Engineering and Semantic WebOctober 7-9, 2012. Saint-Petersburg, Russia (in Russian).

[24] S. Staab, R. Studer, eds. Handbook on ontologies, Springer, 2009.[25] T. Gavrilova Ontological engineering for practical knowledge

work, In Knowledge-Based Intelligent Information and EngineeringSystems, Lecture Notes in Computer Science, Vol. 4693, 2007, pp1154-1161.

[26] R. France, B. Rumpe. "Domain specific modeling." Software andSystems Modeling, vol. 4, no. 1, 2005, pp. 1-3.

[27] D. Koznov, "Process Model of DSM Solution Development andEvolution for Small and Medium-Sized Software Companies." InEnterprise Distributed Object Computing Conference Workshops(EDOCW), 2011 15th IEEE International, pp. 85-92. IEEE, 2011.

[28] D. Kudryavtsev, L. Grigoriev, Systemic Approach TowardsEnterprise Functional Decomposition, Proc. IEEE 13th Conferenceon Commerce and Enterprise Computing (CEC), 2011, pp. 310-317.

[29] Y. Akao (ed.) Hoshin Kanri, Policy Deployment for SuccessfulTQM, Cambridge, MA: Productivity Press, Inc., 1991.

[30] M. Clargo, The designer organisation: Organisations too can benefitfrom the application of design and quality tools, and with startlingresults! International Journal of Quality & Reliability Management,21(9), 2004, pp. 973-983.

[31] T. R. Browning, "Process integration using the design structurematrix." Systems Engineering, vol. 5, no. 3, 2002, pp. 180-193.
http://www.sciencedirect.com/science/journal/09505849http://www.sciencedirect.com/science/journal/09505849/36/6http://publib.boulder.ibm.com/infocenter/rsysarch/v11/topic/com.ibm.sa.help.doc/topics/c_Matrix_Editor.htmlhttp://publib.boulder.ibm.com/infocenter/rsysarch/v11/topic/com.ibm.sa.help.doc/topics/c_Matrix_Editor.htmlhttp://www.casewise.com/products/modelerhttp://www.metacase.com/mep/matrix_editor.htmlhttp://www.metacase.com/mep/matrix_editor.htmlhttp://www.casewise.com/products/modelerhttp://publib.boulder.ibm.com/infocenter/rsysarch/v11/topic/com.ibm.sa.help.doc/topics/c_Matrix_Editor.htmlhttp://publib.boulder.ibm.com/infocenter/rsysarch/v11/topic/com.ibm.sa.help.doc/topics/c_Matrix_Editor.htmlhttp://www.sciencedirect.com/science/journal/09505849/36/6http://www.sciencedirect.com/science/journal/09505849