Topic 15: Object-oriented Design I

Synopsis

• Object-oriented Design
   
• Concepts from the Object Model
   
• The tasks in an object-oriented design
   
• Finding the objects (informal description)
   
• Finding the objects (domain analysis)
   
• Finding the objects (checking)
   
• Finding the classes^¤
   
• Classification^¤ Techniques
   
• Classical classification^¤
   
• A priori (conceptual) clustering
   
• Prototype theory
   
• Analogical approaches
   
• Reading

Object-oriented Design

• Produces a set of class^¤ hierarchies and and specification of class^¤ interactions which solve the specified problem
   
• Cellular analogy
   
• Build a hierarchical system from a small number of independent standard components
   
• System behaviour^¤ "emerges" from the interactions of those components at various levels
   
• Discovery and invention

Concepts from the Object Model

• System will be "noun-based", not "verb-based"
   
• Abstraction^¤: Collect common structure and behaviour^¤ in more abstract classes^¤
   
• Encapsulation^¤: Isolate internals to reduce interdependence
   
• Aggregation^¤: Build classes^¤ out of simpler classes^¤ (inheritance), build objects out of simpler objects (members^¤)
   
• Polymorphism^¤: Control behaviour^¤ at the highest possible level of abstraction^¤ (let the objects work out the details themselves)

The tasks in an object-oriented design

• Find the objects
   
• Find the classes^¤
   
• Specify the relationships
   
• Design the interactions
   
• Refine the design

Finding the objects (informal description)

• Write a NL description of the problem
   
• Nouns may represent objects
   
• Verbs may represent object behaviours^¤ (methods) or occasionally events to be modelled as objects
   
• Adjectives may indicate specializations^¤ (child classes^¤¤) or independent properties (mixin^¤ classes^¤)

Finding the objects (domain analysis)

• Examine the existing system (automated or manual)
   
• What things do the current domain experts talk about when describing it?
   
• Examine similar applications (automated or manual)

Finding the objects (checking)

• Does this thing have a well-defined role in the problem domain?
   
• Does this thing have well-defined boundaries in the problem domain?
   
• Does this thing have a well-defined behaviour^¤ in the problem domain?
   
• If any answer is "no", it's probably not a single object
   
• Example: engine, car, traffic
   
• Example: client, clients, clientele

Finding the classes^¤

• Classification^¤ is hard - "No silver bullet" (Brooks)
   
• Discovery vs Invention
   
• Iterative
   
• Restructuring operations: abstraction^¤, factorization, composition

Classification^¤ Techniques

• Classical (Platonic)
   
• A priori clustering
   
• Prototype theory
   
• Analogical approaches

Classical classification^¤

• Cluster according to important common properties (Plato)
   
• ..and/or according to common effects (Aquinas)
   
• Also known as emergent classification^¤ (classes^¤ emerge from the collection of objects)
   
• Works well if the important common properties used are significant in the domain
   
• "Bottom-up" approach
   
• Example: early zoological classification^¤

A priori (conceptual) clustering

• Formulate the concepts which distinguish objects first
   
• Match objects to each concept to classify
   
• Works well if the concepts selected are significant in the domain
   
• "Top-down" approach
   
• Example: Racism

Prototype theory

• Select a "paradigm" ("perfect example") object for each potential class^¤
   
• Cluster objects which are like the paradigm object into the same class^¤
   
• Works well if the example object really is representative.
   
• "Instance^¤-outwards" approach
   
• Example: Exam marking

Analogical approaches

• Find an existing system which is a strong analogue for the proposed system
   
• Generate classes^¤ for the new system by "reversing the analogy"
   
• Works well if the analogy works well (i.e. depends on creativity and experience)
   
• "Concept-backwards" approach
   
• Example: Ant roles

Reading
 

• Pressman: Chapters 20 - 22
   
• Sommerville: Chapter 12
   
• Stroustrup: Chapters 23 to 25
  
• Schach: Chapters 12 and 13