Theory and Practice

1
Business Domain Modelling Principles
HYPERCUBE Ltd 7 CURTAIN RD, LONDON EC2A 3LT

• Theory and Practice

Mike Bennett, Hypercube Ltd. www.hypercube.co.uk
2
Overview

• Business requirements and semantics
• Business Semantics Theory
• Development process grounding
• Philosophical grounding
• Implementing for Financial services
• Recommendations

3
Top Down DevelopmentSummary of Stages
Analysis Stage Some Model Types Language
1. Business Analysis Business Use Cases, Process models, Business Data Semantics UML
2. Requirements Analysis Requirements Use Cases Business data models Message choreographies UML
3. Logical Analysis Package and Class models Message Sequence Diagrams UML
4. Logical Design Class models UML
5. Physical Design Program code, XML Schema Target language
4
UML based top down development

• Requires business semantics to be modelled in UML
• Unextended UML does not model business meaning
• Extensions can be added but these are not in
  basic UML
• Add own extensions
• Use standard extensions for business meaning

5
Going beyond UML

• Zachman Framework
• Structural and Behavioural aspects of the
  development project
• Corresponding structural and behavioural aspects
  of the business problem domain
• This has a spot on it for business semantics of
  data
• This is where the theory and practice of
  ontologies and taxonomies comes in

6
(No Transcript)
7
(No Transcript)
8
(No Transcript)
9
Top Down Development and Model Driven Architecture

• Not the same thing, but similar disciplines apply
• Decomposition of problem into solution
• Linkages between business (requirements) and
  Developed material
• Business Requirements
• Structural data requirements
• Behavioural business process

10
MDA Framework Deliverables

• Structural aspect (business data requirements)
• Model business Terms, Definitions and
  Relationships (TDR)
• Use Ontology approach
• Behavioural aspect (business process)
• High level requirements for business parties to
  do business
• Decomposes into message choreographies
• Link this to ontology

11
Philosophical Foundations

• We must philosophise if only to avoid
  philosophising Aristotle
• We must deal with ontology if only to avoid
  ontological mishaps Bennett

12
Philosophical Foundations

• There are three considerations in computer
  knowledge representation (Sowa, 1995)
• Logic
• Ontology
• Computation

13
The Ancient Philosophers

• Aristotle
• Knowledge representation
• Logic Representation
• Invented the Variable
• Leibniz, Kant, Hegel, Russel Whitehead etc.
• Logic Representation
• Peirce, Whitehead etc.
• Ontology

14
Definitions

• Taxonomy
• Hierarchical categorisation of concepts in the
  domain of discourse
• Ontology
• Adds descriptions, properties and restrictions to
  the concepts in a taxonomy
• The properties include relationships among
  concepts
• These definitions are based on the best available
  from academia.

15
What is an Ontology?

• an ontology is a model which has
• Formal explicit description of concepts in a
  domain of discourse (referred to as Classes)
• Properties of each concept (class) describing
  features and attributes (known variously as
  slots, properties or roles)
• Restrictions on those properties (known as
  facets).

16
Ontology Principles

• Set theory - determine what object, from the set
  of all possible things, is in this class
• Uses Predicate logic to define this
• some ontology tools use the original Predicate
  Calculus notations (Russell et al)
• some tools represent this in natural language

17
Ontology and Taxonomy

• Provides the predicates for logic to operate on
• Ontological commitment
• Deciding what will be represented in your
  ontology
• This will be a taxonomic hierarchy of Classes
  with Generalisation (OO Inheritance)
  relationships
• can be represented in UML or in RDFS (Taxonomy)
  or OWL (ontology)

18
Putting Ontology into Practice

• Ontological commitment what goes into the
  taxonomy
• Need a Taxonomy before you can do an ontology,
• BUT
• the classes in the Taxonomy must be ones that
  would make sense in a complete ontology

19
Creating an Ontology

• Define the Taxonomy
• in RDFS (standard taxonomy format)
• or create in UML using inheritance only, and
  import into Ontology tool
• Promote Taxonomy classes to ontology hierarchy
• commit the relevant classes to be sub-classes of
  owlThing in the ontology tool

20
Ontology in Practice What you will see in an
ontology tool

• Namespaces
• Classes of Thing
• Generalisation relationships
• Disjoints
• Properties
• Object Property
• Attribute Property
• Annotation Property
• Property Characteristics
• Adds more semantics to the properties

21
What is a Thing?

• Financial Instrument
• Issuance Terms (a contractual kind of thing)
• Cash Flows
• Equity
• Debt
• (are these sub-types of a general kind of thing?)

22
Possible Classes of Thing
23
Expanding the Taxonomy
24
What is an Equity?

• "An equity is a financial instrument setting out
  a number of terms which define rights and
  benefits to the holder in relation to their
  holding a portion of the equity within the
  issuing company".

25
What is an Equity?
Or to put it another way
Equity
Financial Instrument
Instrument Terms
In relation to
Is a kind of
Has rights defined in
Equity security
26
What is an Equity?
27
Developing the Ontology

• We can expand on the above to define more
  information about our reality, such as the
  characteristics of different classes of equity
• Each attribute and relationship adds information
  to define what qualifies a thing as uniquely
  belonging to a given class
• Ontology uses set theory and predicate logic to
  determine what is in what class.

28
Notes on these views

• The above views are taken from the tool TopBraid
  Composer
• This uses the OWL language but gives a more
  graphical representation than other tools such as
  Protégé
• The Predicate Calculus is hidden away under the
  hood but it is still there.

29
Extending UML

• Can extend UML to support OWL
• OMG Recommendation
• OWL Extensions
• The same principles of taxonomy and ontology
  apply
• Can be consumed by MDA efforts
• Will not have all the powers of an ontology tool
• use for representation not for processing

30
Recommendations

• Use OWL Ontology language for business semantics
• Use RDFS to capture Taxonomy initially
• Need guidelines for identifying taxonomic
  categories
• Extend to OWL as required
• Either in extended UML or native OWL
• Determine best business review format

31
Standards Based Taxonomies

• Some standards provide a true taxonomy already
• e.g. ISO 10962 (Classification of Financial
  Instruments)
• should not be duplicated by parallel structures
  in your own model
• should plug in to a Taxonomy as a distinct
  category of Thing
• Other standards do not

32
Summary Principles

• Representing business knowledge requires a
  different approach from representing data
• Define a taxonomic hierarchy of separate
  categories of real-world Thing at the outset
• Relationships can be added to define an ontology
  of the business domain
• As long as they only have business meaning
• Define the reality ahead of any design or
  development
• Good development weak semantics
• Dont try and use a technology format (UML, XML)
  to represent business semantics. Use the best of
  breed for each requirement.

33
Tools

• Protégé
• Freeware
• Good for logical statements
• TopBraid Composer
• Open Source and inexpensive
• Good graphical views
• UML Editors (MagicDraw, EA)
• Can be extended to represent OWL
• Easier to link into MDA framework
• But would not have all the powers of an OWL tool

34
Questions?

• Mike Bennett
• Hypercube Ltd.
• 7 Curtain Road, London EC2A 3LT
• mbennett_at_hypercube.co.uk
• 020 7917 9522
• www.hypercube.co.uk