Chapter 6 Models

1
Chapter 6Models

• What are models?
• Models capture the essential aspects of a system
• Models ignore nonessential aspects.
• Can serve as a blueprint for new systems.
• Used to evaluate existing systems.
• Have to be expressed clearly and concisely
• Mathematical notation is ideal
• Not possible to give a single comprehensive model
  for any given system
• Examples ANSI/SPARC model, and ISO OSI model

2
Chapter 6Models

• Model progression
• The nature of the model depends on the maturity
  of the research problem.
• Often a research opportunity (problem) appears.
• The precise nature of the problem is not clear.
• A tentative model helps to clarify the problem.
• After a couple of clarification models, models
  are produced to analyze the problem.

3
Chapter 6Models

• At this point it is necessary to look at the
  assumptions for generating the models in the
  first place, and whether those assumptions are
  actually valid, and whether alternative
  assumptions can be established.
• If different assumptions are made, new models
  have to be formulated, referred to as
  differentiated models.
• Trends are observed and general models are
  proposed. They usually cater for most of the
  earlier assumptions.

4
Chapter 6Models

• Clarification stage
• Sometimes a problem that needs a model to address
  it simply appears.
• New application previously not thought possible
• Researchers have to actively search for model
  problems.
• Often new paradigm requires new solutions for
  problems.
• Often such models require new models for all
  existing problems known for the old model.
• Transactions, concurrency, management etc.

5
Chapter 6Models

• Differentiation stage
• Once the 1st models have been proposed, the
  merits for investigating the problem have been
  established.
• Determine now whether the models are adequate, or
  only form the starting point.
• Investigate the assumptions, starting points and
  restrictions of already proposed models
• Propose a model that differs from existing models

6
Chapter 6Models

• Indicate why your model is better, or why it is
  needed in addition to existing models
• It may address cases not addressed by original
  model
• May provide a better fit for new paradigm
• May be simpler
• May employ mathematical notation (more exact)
• May be merely different, therefore more acceptable

7
Chapter 6Models

• Generalisation
• General models are surprisingly simple
• Can be described in essence using a simple
  diagram and lines of text
• Details are hidden behind this simple exterior

8
Chapter 6Models

• How to model
• Work, insight and often luck.
• Approaches
• By design
• Design of systems is common to IT.
• Common methods include
• Identify the major components of a software
  system
• Use these components as the components of the
  model
• Use dataflow techniques to identify major events
  of a process
• Use these techniques to construct a model.

9
Chapter 6Models

• By metaphor
• Much done is analogous to the real world.
• Concepts are derived from the real world (files,
  windows etc)
• Suitable names are easy to remember
• Acceptable names bring new possibilities with
  them
• Ideas carried over from the metaphor often enrich
  the concept
• Opens possibilities that may have otherwise never
  been considered
• How to find a metaphor
• Know the essence of what you want to express
• Think creatively
• Look at your environment actively with a
  receptive mind

10
Chapter 6Models

• By formalisation
• Is usually relatively straightforward
• Choose your tools carefully
• Tools should be most applicable for the sorts of
  manipulation that you envisage
• Viable alternatives include
• Set theory
• Logic
• Abstract algebra
• Formal languages
• Automata theory

11
Chapter 6Models

• By pure genius
• It happens sometimes

12
Chapter 6Models

• Why use Models?
• 1. Simplicity
• Millers rule humans can only have seven
  concepts in short term memory at a time
• Models make it possible to comprehend the essence
  of the modeled concept
• 2. Comprehensiveness
• Models often systematically address all aspects
  of a problem
• Prevents you from getting lost in the detail of
  the actual problem
• The more aspects of a problem your model covers,
  the better

13
Chapter 6Models

• 3. Generality
• The more variations your model addresses the
  better
• 4. Exactness
• If the model closely fits the perceived problem,
  it is more likely to be accepted
• E.g. if a metaphor is used, the perceived
  applicability of the metaphor helps to determine
  its acceptability

14
Chapter 6Models

• 5. Clarity
• Purpose of all components should be evident
• Operation of each facet should be evident
• Interaction or flow between components should be
  evident
• Guard against ambiguity
• Ask what the major strengths of the model are
• What are the strengths compared to existing
  models?
• What are its intrinsic strengths?
• Point out the strengths to your audience