IS5411 Chap.7 Domain Classes and Relationship Models
Domain Classes and Relationship Models
The noun technique starts a decision process
Nouns from use cases, actors, inputs, outputs and documents are candidate classes, not automatic tables. Include a candidate when it is a distinct in-scope thing with multiple instances and meaningful attributes or relationships that the system must remember.
Exclude synonyms, actions, outputs, examples, the system itself and values better treated as attributes.
Relationships must preserve business meaning
Ordinary associations connect participating things. An association class stores facts belonging to a many-to-many pairing. Whole-part relationships express containment with an explicit lifetime rule.
Generalisation and specialisation express a genuine kind-of relationship and inheritance. Visual importance never justifies a stronger relationship symbol.
Multiplicity is a directional claim
Read every association in both directions before placing symbols. A loan can belong to one user while a user accumulates many loans over time. Optionality belongs where the business rule permits no related instance.
Each multiplicity must trace to requirements evidence rather than to a familiar database pattern.
What this chapter covers
- 01
Noun extraction
- 02
Include decisions
- 03
Exclude decisions
- 04
Attributes and types
- 05
Association classes
- 06
Whole-part relationships
- 07
Inheritance
- 08
Multiplicity audit
Worked example · free
Model workshop enrolment
- 3Identify the many-to-many relationship.
- 4Justify the association class and attributes.
- 3State both directional multiplicities.
Key terms
- Noun Technique
- A method that extracts candidate things and classifies them by scope, memory and relationships.
- Domain Class
- A distinct problem-domain thing about which the system preserves information.
- Attribute
- A piece of information describing an entity rather than an independent remembered thing.
- Association Class
- A class holding information that belongs to a many-to-many relationship.
- Composition
- A whole-part relationship in which the part depends on the whole for identity or lifetime.
- Generalisation
- A hierarchy in which a specialised subclass inherits characteristics from a broader superclass.
Domain Classes and Relationship Models FAQ
Should every noun become a class?
No. Nouns are candidates that must pass the memory test. Exclude the system, interfaces, actions, outputs, synonyms, examples and values better represented as attributes. Include distinct in-scope things with repeated instances and their own meaningful attributes or relationships.
When is an association class necessary?
Use one when a many-to-many relationship has information of its own, such as a grade for one learner-section pairing or a status for one enrolment. The class preserves facts that belong to neither endpoint alone and gives the relationship a business identity.
How do aggregation and composition differ?
Both are whole-part relationships, but composition expresses a stronger lifetime dependency between part and whole. Do not use the filled diamond to show importance. Select the relationship only when the domain states whether parts can exist independently and be moved or replaced. Describe the lifetime rule in words before selecting either diamond notation.
Exam move
Extract nouns from a short requirement and create an include, exclude or research table with a one-sentence reason for every decision. Draw relationships only after the class list stabilises. For each line, state the relationship and multiplicity in both directions, then check whether the requirements actually support the sentence. Run four separate audits. First, remove the system, interfaces, actions and outputs.
Second, collapse synonyms and decide whether examples or types require distinct structure. Third, test whether apparent entities are attributes by asking whether they have their own repeated instances, attributes or relationships. Fourth, recover transaction classes whose dates, statuses or links belong to a pairing rather than to either endpoint. Only then add association, whole-part or generalisation relationships.
Defend composition with a lifetime dependency and generalisation with a genuine kind-of rule. Read optionality aloud from both ends, including the zero case. Revisit the use-case outcomes at the end: every durable result should have a place in the model, and every added class should trace back to a requirement rather than database habit.
Redraw the model after removing every relationship symbol, then restore each one from a sentence in the requirements. This deliberately slow pass distinguishes remembered business rules from familiar UML shapes and is especially effective for association classes and whole-part claims.
Working through Domain Classes and Relationship Models in IS5411? Sia is AskSia’s AI Information Technology tutor — ask any IS5411 Domain Classes and Relationship Models question and get a clear, step-by-step explanation grounded in how IS5411 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.