SEMANTICS AND PRAGMATICS IN ENTERPRISE
ARCHITECTURE THROUGH TRANSACTION
AGENT MODELLING
Ivan Launders, Simon Polovina
Conceptual Structures Research Group, Sheffield Hallam University, Arundel Street, Sheffield, U.K.
Richard Hill
Distributed and Intelligent Systems Research Group, University of Derby, Derby, U.K.
Keywords: Semantics, pragmatics, Enterprise architecture, Transaction agent modelling.
Abstract: Enterprise architectures comprise of complex transactional information systems that perform repetitive and
bespoke business transactions to meet business goals. Frameworks for enterprise architectures have been
widely adopted to organise design thinking about the architectural components as well as to provide a
description of architecture artefacts. We note various shortcomings of these framework approaches, giving
rise to how semantics and pragmatics should evolve in enterprise architectures through Transaction Agent
Modelling (TrAM). We accordingly outline steps for capturing and modelling the semantics in business
transactions for enterprise architecture.
1 INTRODUCTION
Frameworks for enterprise architectures have been
widely adopted to help organise design thinking
about complex transactional information systems
(TOGAF 2009, and Zachman 1987). Sowa &
Zachman provided an original vision of what a
contemporary enterprise architecture should be
(Sowa and Zachman 1992), extending the vision of
an Information System Architecture (ISA)
framework to show how it could be formalised in the
notation of Conceptual Graphs (Sowa 1984).
Zachman recognised that modelling tools and
techniques of the day (e.g. entity-relationship
diagrams, object-oriented systems, UML) were
specialised for different purposes and that by
concentrating on one aspect, a technique can lose
sight of the overall information system and how it
relates to the enterprise (Fowler 2004, Jacobson et
al. 1992, Sowa and Zachman 1992). Sowa and
Zachman’s approach to Information System
Architecture (ISA) was to use a framework to
provide a system architecture scope; an enterprise or
business model; a system model, and finally a
technology model and its subsequent model
components (Sowa and Zachman 1992). Today an
ISA would be referred to as an Enterprise System
Framework.
Architecture development frameworks provide a
structure within which the key components of the
architecture and the relationship between those
components are defined (Sowa and Zachman 1992,
TOGAF 2009, Zachman 1987). A framework helps
to organise our thinking about software architecture
and provides a description of artefacts. It helps
ensure that the semantics in an enterprise are widely
understood. Attributes of a good architecture
development framework include:
- consistency and structure;
- capturing the 'kite' or high level process;
- incorporating a variety of constructs at different
levels of abstraction;
- a defined, enabling process for developing the
architecture;
- a description of the artefacts that will be
produced during the work of the architecture
development;
- a clearly described process.
The Open Group Architecture Framework
(TOGAF) provides a significant move towards
285
Launders I., Hill R. and Polovina S.
SEMANTICS AND PRAGMATICS IN ENTERPRISE ARCHITECTURE THROUGH TRANSACTION AGENT MODELLING.
DOI: 10.5220/0003268302850291
In Proceedings of the Twelfth International Conference on Informatics and Semiotics in Organisations (ICISO 2010), page
ISBN: 978-989-8425-26-3
Copyright
c
2010 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
Figure 1: TOGAF ADM with added semantics
working with a formal Enterprise Architecture
framework, by the development of a meta-
Framework (TOGAF 2009) that is designed to be
customised and could be extended to add semantics.
TOGAF was first developed in 1995 based on the
US Department of Defence Technical Architecture.
It provides a tool for assisting the acceptance,
production, use and maintenance of architectures.
The approach is an iterative process model focusing
on architecture types including:
- Business Architecture: The business strategy,
governance, organisation and the key business
processes. Semantics would enhance this
architecture type.
- Data Architecture: The structure of an
organisation's logical and physical data assets
and data management resources.
- Application Architecture: The blueprint for
applications, and their relationships to the core
business processes of the organisation.
The Data Architecture and Application
Architecture comprise the Information Systems
Architecture of TOGAF.
Capturing a business process “As Is” provides a
record of the business as it currently works; in
TOGAF this is referred to as a “baseline”. Analysing
enterprise business processes with high level
business goals, such as improving efficiencies and
performance, and then having the ability to be able
to model, automate, and visualise those potential
improvements, offers business transactions
refinement at a semantic level. Transaction Agent
Modelling (TrAM) extends requirements capture
using the rigour of Conceptual Graphs (CG) and
Resource-Events-Accounting, latterly referred to as
the Resource-Event-Agent (REA) model (Geerts and
McCarthy 1991, McCarthy 1982, McCarthy 1979,
and Polovina 1993).
We outline an automated approach to assist
designers with the capture of semantics in enterprise
transactions through the use of a generic Transaction
Model (TM), allowing for business transactions to
be enriched and refined at the early stage in the
design process. We explain how an improved
framework that places greater emphasis upon the
capture of semantics in business transactions
through the automation of CG can assist an
enterprise architect (Hill and Polovina 2008,
Polovina 2007, Polovina and Hill 2009, and
Launders et al. 2009).
2 SEMANTICS AND
PRAGMATICS IN BUSINESS
ARCHITECTURE
The linguistic understanding of the word
“semantics” is the study of meaning itself (Sowa
1984). Pragmatics studies how the basic meaning is
ICISO 2010 - International Conference on Informatics and Semiotics in Organisations
286
related to the current context and the listener’s
expectations. Syntax studies the grammar rules for
expressing meaning in a string of words (Sowa
1984, Stamper 1996). Semantics determines the
literal meaning. Other factors, which relate language
to the world, are called pragmatics. In the context of
achieving successful communications in an
enterprise, stake-holders in business transactions
need to be consistent in the use of language. We
focus attention on where the semantics are in
business transactions and how we capture and work
with them in enterprise architecture. The Semantics
of Business Vocabulary and Business Rules (SBVR,
http://www.businessrulesgroup.org/sbvr.shtml)
group provides an approach that enables people and
organisations to treat business, legal, and
educational knowledge in a productive and
consistent way. SBVR aims to combine the valuable
aspects of logic, natural language, business rules,
and conceptual modelling. Business knowledge can
be described using:
- a business concepts catalogue;
- an association of business fact types;
- a business rules catalogue.
A business concepts catalogue is identified by
SBVR, as a major step forward, in that consistency
in the use of concepts in business rules is important
in ensuring the quality of business rules. A
description should be consistent with Sowa’s
definition, that a conceptual catalogue shows how
the form can be applied to the words and concepts
(Sowa 1984).
One of the areas of future development in
architectural frameworks such as TOGAF could be
in the capture of semantics in the business
architecture. Figure 1 provides the Architectural
Development Method (ADM) for TOGAF
illustrating where to add semantics into the ADM
phases. The Business Architecture phase 'B' could be
extended to capture the semantics in high level
business transactions.
A TOGAF Architecture Vision as shown in
Figure 1 starts by articulating business requirements
implied in new business functionality to meet
business goals. It then implies a technical
architecture requirement. Two key elements of this
step include identifying:
- Human Actors: Identify human actors and their
place in the business model, the human
participants and their roles.
- Computer Actors: Identify computer actors and
their place in the technology model, the
computing elements and their roles.
TOGAF states that a variety of modelling tools
and techniques may be employed, if deemed
appropriate, therefore we have considered a
Transaction Agent Modelling framework referred to
as TrAM (Hill 2005). TrAM is a requirements
elicitation framework for agent-oriented software.
Activity Models (also called Business Process
Models) describe the information exchange
functions (internal and external) associated with the
enterprise's business activities, from a data
perspective. Activity Models are hierarchical in
nature, capturing the activities performed in a
business process, and TOGAF refers to these as
ICOMs (inputs, controls, outputs, and mechanisms/
resources used) of those activities. Models
represented as use case diagrams can describe either
business processes or systems functions, depending
upon the focus of the modelling effort. In TrAM we
refer to Transactional Use Cases (TUC) which are
specifically focused on enterprise transactions. TUC
describe at a high level the primary business
transactions in an enterprise in terms of use cases
and internal and external actors, corresponding to
business transactions. TrAM provides a greater level
of focus, capturing semantics within the business
transaction initially captured in TUC, thus
developing upon TOGAF’s more general approach
to activity and use case modelling.
An experienced software designer would
consider enterprise architecture from a human and
social level but they would tend to deal with it
intuitively and informally using experience and
domain knowledge from previous work (Stamper
2007). CG allow the designer to represent
knowledge and to use interactive tools such as
Amine (http://amine-platform.sourceforge.net/) for
representing natural language in a structured,
knowledge representation language. Stamper’s
“Semiotic Ladder” helps identify that information
systems function adequately when signs are handled
correctly on every architectural level, and that
ineffective systems tend to ignore problems on one
or more levels, typically the upper three levels of the
Semiotic Ladder (social world, pragmatics and
semantics) (Stamper 1996, Stamper 2007).
Technical architecture levels are often more
accurately specified and designed, whereas this is
more unlikely in the analysis and design of
semantics and pragmatics in enterprise architecture.
Conceptual analysis enables software designers
to define the schemata for enterprise architecture.
Applying conceptual analysis through TrAM
provides the focus upon an enterprise transaction
through the rigour of CG by providing model
SEMANTICS AND PRAGMATICS IN ENTERPRISE ARCHITECTURE THROUGH TRANSACTION AGENT
MODELLING
287
checking to assist in the early requirements capture
(Hill et al. 2005). The results of the analysis and the
description of the concepts within a particular
domain are known as an ontology (Gruber 1993). An
ontology contains the understanding of knowledge
in a given domain and creating that ontology
requires input from the human information functions
(semantics, pragmatics, and social world) through
domain knowledge. Creating that ontology
containing enterprise semantics can significantly
benefit from the CG approach particularly in
conjunction with software tools for the development
of intelligent systems. This process is likely to be
iterative.
3 SEMANTICS AND
TRANSACTION AGENT
MODELLING
The main intent of semantics in communications is
to give machines better access to information so they
can be information intermediaries in support of
humans (Burners-Lee et al. 2001). When agents
communicate with each other, there needs to be a
means of exchanging meaning through a transaction,
so agent A has the same conceptual understanding as
agent B in a business transaction. If we accept that
agents may not use the same terms to mean the same
things, we need a way to discover what another
agent means when it transacts (Uschold 2001). In
order for transactions to happen, every agent will
need to declare or catalogue what terms it is using
and what they mean, much the same as human
agents do linguistically through a glossary. This
specification is referred to as the agent’s ontology
(Gruber 1993). An ontology can describe meaning
as a formal specification of the terms in a domain
and the relationships between them, using a common
vocabulary for agents who need to share information
in that domain.
Figure 2: Software Agent Encoding.
Semantics must be accessible to software agents
and therefore need encoding in some form of logic
(Uschold 2001), as illustrated in the community care
example in Figure 2. This will enable software
agents to use automated reasoning to accurately
determine the meaning of other software agents. In
practice there are many difficulties to overcome for
this to happen reliably and consistently. However
before the stage of ontology representation a
designer needs to capture and model the semantics
in business transactions. Whilst enterprise
architecture modelling can facilitate the visualisation
of business processes, there are limitations,
namely: the accuracy of the process model; the
completeness of a model, and therefore being aware
of what is missing; if there are any efficiencies to be
had which an enterprise could benefit from.
In the context of semantics TrAM focuses on the
machine processing semantics for REA type
business transactions. The approach TrAM takes is
to capture and represent these semantics, in order
that a base ontology can be derived, but also to
support the reasoning of new concepts as and when
they occur. TrAM theory defines that enterprise
system designs will include the following:
- Model fundamentals: Providing a complete
Transactional Use Case diagram (TUC)
capturing the transactional behaviour of the
initial use case. A close mapping between TUC
and CG, translating that transactional behaviour
and adding semantics through comprehensive
analysis with CG, including co-referent links and
a supporting type hierarchy (Hill et al. 2005).
- Model Visualisation: Visualisation of business
rules, using Peirce logic. CG are a system of
logic based on Peirce’s Existential Graphs
(Roberts 1973). Proof of the enterprise business
rules, specialisation and projection within the
proof of the design rules.
- Model Automation: CG tools such as Amine
(http://amine-platform.sourceforge.net/) provide
good use of transferring the initial transaction
model analysis into an Amine ontology with
accurate type hierarchy and TM. Integration of
the model with a conceptual catalogue (CC)
showing how form can be applied to words and
concepts in the transaction model. Use of the
ontology to achieve a successful projection with
the inclusion of the business rules in the
ontology.
ICISO 2010 - International Conference on Informatics and Semiotics in Organisations
288
3.1 TrAM Model Automation
Model automation involves building the ontology
for the TM as well as integrating a Conceptual
Catalogue into the ontology to describe the form and
meaning of the concepts in the model (Hill and
Polovina 2008, Uschold 2001, Launders et al. 2009).
Figure 3: TrAM, with automation.
Figure 3 provides an illustration of TrAM with
automation, where the designer transforms a paper
based CG analysis (Model Fundamentals & Model
Visualisation) into a software model for verification
(Model Automation) through CG operations using
Amine. The steps for automation expand upon Hill's
original TrAM framework (Hill et al. 2005). These
steps are as follows:
- Model Fundamentals
1. Capture Transactional Use Case logic (TUC)
2. Create a Transaction Model
a) Transform TUC into CG
b) Integrate CG with the generic TM
c) Iterate TM with TUC
- Model Visualisation
3. Visualise the TM using Peirce Logic to
highlight any further requirements.
- Model Automation
4. Create a Conceptual Catalogue for the
transactional terms used in the TM
5. Verify the semantics and logic captured in
the TM (Inference against models and verify
using Amine)
a) Create and Refine the Type Hierarchy;
b) Build the Ontology for the TM (or
Ontologies if there are more than one)
c) Integrate the Conceptual Catalogue
6. Specialise the TM through business rules
(Test and refine the TM with business rules
using Amine).
The ‘transactional use case’ (TUC) diagram
captures high level transaction logic, to be
transferred into Transaction Model Conceptual
Graphs, representing concepts and relations for each
of the high level transactions identified. The TUC
capture involves analysing key transactional facts
from the case study narrative. Central to this analysis
and analogous to Zachman is identifying the ‘What’
(Economic Resources?), ‘How’ (Economic
Events?), ‘Who’ (agents?), and ‘Why’ (business
goals).The subsequent step involves mapping those
use cases into CG, translating and adding semantics
through the use of a conceptual catalogue and Model
Verification. The resulting Transaction Model (TM)
ontology is then tested and refined specialising the
transaction logic through the application of business
rules. The resulting design artefacts include:
- a refined Transaction Use Case (TUC);
- a transformation of TUC into CG;
- a Transaction Model in CG consisting of a type
hierarchy together with domain constraint rules
modelled with Peirce Logic;
- Conceptual Catalogue of domain specific terms,
automated and re-usable;
- an automated TM Ontology which has been
tested and refined through CG operations using
Amine.
These artefacts result in a design specification
for the eventual enterprise architecture that does not
impose a particular implementation and serves to
complement architecture framework that lack a
semantic requirements gathering stage such as
TOGAF.
Figure 4: TrAM Semantics and Pragmatics.
Figure 4 illustrates the design artefacts in layers
showing how they sit in relation to each other and
where semantics and pragmatics are modelled. The
ontology layer through a conceptual catalogue (CC)
describes the form to apply to words and concepts,
as canons and definitions in capturing the semantics
(http://cg.huminf.aau.dk/Module_III/1152.html).
The transaction model (TM) layer following a
generic TM uses the CC and ontology capturing the
negotiation between events and resources. A top
layer including the business rules specialises the
enterprise transaction.
SEMANTICS AND PRAGMATICS IN ENTERPRISE ARCHITECTURE THROUGH TRANSACTION AGENT
MODELLING
289
4 SIMULATION OF INDUSTRIAL
PRACTICE
Our experience has been informed through an
innovative approach to LTA (learning, Teaching and
Assessment), where student design teams experience
the application of this emerging computing theory in
architecture for enterprise applications. We have
used a combination of case studies both ‘real world’
and ‘fictitious’ as a simulation of industrial practice,
for example a mobile NHS case study allowing
clinicians to be able to access patient records during
visits to patients is a direct capture of an industrial
experience. Each case study provided business
transactions with different business settings. One of
the goals of analysing design data from multiple
case studies (or cross case) is to examine if the
events and process on one well described setting can
occur in a different setting. Each case study
contained a narrative account of a situation focusing
on the business transactions and exchange of
resources for events.
5 CONCLUDING REMARKS
Complex enterprise transactions need framework
approaches which create deeper understanding of the
semantics in a business domain. A semiotic
perspective offers a route to deeper understanding,
and an emphasis on the sign and communications
character of information systems (Goldkuhl and
Agerfalk 2002, Stamper 1996, Stamper 2007). Early
requirements modelling helps examine the concepts
and semantics in transactions and therefore visualise
and deepen the understanding of how the business
will actually perform under specific conditions.
Visual models are important for clarifying the
meaning in business transactions at different levels
of abstraction. However, abstract models have their
limitations such as how closely they actually relate
to an enterprise and therefore how complete they are
in terms of capturing semantics and visualising
possible efficiencies. We propose therefore, that
through our illustration of integrating the semantics
(and semiotics) of TrAM, with the development of
CG automation tools such as Amine, together with
TOGAF, addresses the sufficient complexity of the
real world in which businesses operate.
REFERENCES
Berners-Lee, T., Hendler, J., Lassila, O. (2001) “The
Semantic Web”, Scientific American, Vol. 284, Issue
5.
Fowler. M. (2004) UML Distilled (Third Edition),
Addison-Wesley, 103-104.
Geerts, G. L., McCarthy. W. E. (1991). “Database
Accounting Systems”, in Information Technology
Perspectives in Accounting: an Integrated Approach,
Chapman and Hall, 159-183.
Goldkuhl, G. Agerfalk, P. J. (2002) “Actability: A Way To
Understand Information Systems Pragmatics“, In Liu,
K et al. (eds.), Coordination and Communication
Using Signs: Studies in Organisational Semiotics – 2,
Kluwer Academic Publishers, Boston, 2002
Gruber, T.R. (1993). A Transaction Approach to Portable
Ontology Specification. Knowledge Acquisition 5:
p199-220
Hill, R., Polovina, S., (2008) “An Automated Conceptual
Catalogue for the Enterprise”, Supplementary
Proceedings of 16
th
International Conference on
Conceptual Structures (ICCS 09): Conceptual
Structures: Knowledge Visualization and Reasoning,
Toulouse, France, July 2008, Eklund, P., Haemmerlé,
O. (Eds.), CEUR-WS, Vol-354, 99-106.
Hill, R., Polovina, S., Beer, M. D. (2005) "From concepts
to agents: Towards a framework for multi-agent
system modelling", Proceedings of the Fourth
International Joint Conference on Autonomous Agents
and Multiagent Systems (AAMAS), Utrecht
University, Netherlands, ACM Press, 1155-1156.
Jacobson, I., Christerson, M., Jonsson, P. and Overgaard,
G. (1992) Object-Oriented Software Engineering.
Wokingham, England: Addison-Wesley.
McCarthy, W. E., (1982) “The REA Accounting Model: A
Generalized Framework for Accounting Systems in a
Shared Data Environment”, The Accounting Review,
554-578.
McCarthy, W. E., (1979) “An Entity-Relationship View of
Accounting Models”, The Accounting Review, 667-
686.
Polovina, S., (2007). "An Introduction to Conceptual
Graphs", Proceedings of the 15th International
Conference on Conceptual Structures (ICCS 2007):
Conceptual Structures: Knowledge Architectures for
Smart Applications, July 2007, Sheffield, UK; Priss,
Uta; Lecture Notes in Artificial Intelligence (LNAI
4604), Springer, 1-15.
Polovina S., Hill, R. (2009) "A Transactions Pattern for
Structuring Unstructured Corporate Information in
Enterprise Applications", International Journal of
Intelligent Information Technologies, April-June
2009, Vol. 5, No. 2, IGI Publishing, 34-48.
Polovina, S. (1993) "Bridging Accounting and Business
Strategic Planning Using Conceptual Graphs",
Conceptual Structures: Theory and Implementation,
Pfeiffer, Heather D; Nagle, T. (Eds.), LNAI, Springer-
Verlag, Berlin, 312-321.
ICISO 2010 - International Conference on Informatics and Semiotics in Organisations
290
Sowa, J. F., (1984). Conceptual Structures: Information
Processing in Mind and Machine, Addison-Wesley.
Sowa, J., Zachman, J.A. (1992) “Extending and
formalizing the framework for information systems
architectures“, IBM systems journal VOL 31, No 3,
1992
Stamper, R. (1996). Signs, Norms, and Information
Systems. In B. Holmqvist et al. (Eds.), Signs at Work.
Berlin, Germany: Walter de Gruyter (pp. 349-397).
Stamper, R. Stumbling across a “soft Mathematics” while
exploring some issues of Organisation, Law and
Metaphysics. Unpublished, Proposed for ICCS 2007.
TOGAF, The Open Group Architecture Framework
(TOGAF), Version 9, Enterprise Edition.
Launders, I. (2009). “Socio-Technical Systems and
Knowledge Representation” Whitworth , B., & De
Moor, A. (Eds.). (2009). Handbook of Research on
Socio-Technical Design and Social Networking
Systems. Hershey; 1,034 pp
Launders, I., Polovina, S., Hill, R., (2009). "Exploring the
Transaction through Automating TrAM", ICCS 2009:
Supplementary Proceedings of the 17th International
Workshops on Conceptual Structures; , Moscow:
Springer
Roberts D. D, (1973) The existential Graphs of Charles S.
Peirce. The Hague, The Netherlands: Mouton.
Uschold, M., (2001). “Where are the Semantics in the
Semantic Web”. Autonomous Agents Conference,
Montreal, June 2001.
Zachman, J.A. “A Framework for Information Systems
Architecture,” IBM Systems Journal 26, No. 3, 276-
292 (1987).
SEMANTICS AND PRAGMATICS IN ENTERPRISE ARCHITECTURE THROUGH TRANSACTION AGENT
MODELLING
291