INF2170H Chap.3 Classification and Faceted Schemas
Classification and Faceted Schemas
Keep one question behind each array
Classification begins with a universe and a characteristic used to divide it. Coordinate classes derived through one characteristic form a homogeneous array, while a chain represents increasing specificity.
This chapter diagnoses mixed sibling lists, compares arrays with chains and uses context, currency, consistency and hospitality as repair checks.
Build independent facets
Facet analysis identifies dimensions that can combine without forcing every compound object into a single branch. The method moves from candidate terms to homogeneous facets, scope notes, relationships and testing.
Metadata rules then state which fields are global, type specific, required, repeatable or controlled.
Repair the relationship before rewriting the label
A familiar list can still mix audience, format, topic and organisational unit. Turn each label into the question it answers. Values that answer different questions belong in separate facets, not one sibling set.
Values at increasing levels of specificity belong in a chain rather than a coordinate array. Once those relations are coherent, choose labels that communicate the intended boundary. Wording alone cannot repair a mixed characteristic.
Context and currency control meaning. A short term can be clear beneath one parent and ambiguous beneath another.
A technically precise label can fail when the intended community uses different current language. Use the parent scope and user vocabulary to choose the smallest clarification that prevents a plausible wrong interpretation.
Keep scope notes with the controlled value so later maintainers do not have to infer the distinction from examples.
Test facets with compound and future objects
A faceted schema should let an object receive independent values without duplication. A training video can have format, audience and topic values simultaneously.
Test whether each facet adds useful distinction, whether combinations support real tasks and whether the current collection merely creates an accidental one to one pattern. If a new object class exposes global fields that do not apply, move those fields to the appropriate type layer.
Hospitality is tested through change. Add a plausible coordinate value and a narrower subordinate value.
The structure should create a distinct place while preserving the meaning of existing classes. Add a borderline object as well. Overlap may require clearer criteria, repeatable values or a broader scoped class. Record the rejected arrangement and repair so future contributors understand why the schema uses several facets instead of returning to one convenient mixed menu.
What this chapter covers
- 01
Universe
- 02
Characteristic
- 03
Homogeneous array
- 04
Class chain
- 05
Facet analysis
- 06
Schema hospitality
Worked example · free
Repair a mixed resource menu
- 3Name the universe and competing characteristics.
- 3Form homogeneous arrays.
- 2Test one compound resource.
Key terms
- Universe
- Universe is the defined set of things under classification.
- Characteristic
- Characteristic is the question or quality used to divide a universe.
- Homogeneous Array
- Homogeneous Array contains coordinate classes derived through one characteristic.
- Class Chain
- Class Chain arranges subordinate classes through increasing specificity.
Classification and Faceted Schemas FAQ
How can a mixed classification be identified?
Ask which question every sibling answers. If one value describes audience, another format and another topic, they are not coordinate classes. Split the characteristics into facets, then define scope notes and test difficult objects before revising labels.
What is the difference between an array and a chain?
An array contains coordinate values derived from the same characteristic, such as several formats. A chain represents progressive subordination, such as a topic and a narrower subtopic. Mixing those levels creates overlapping or uneven siblings.
Why does schema hospitality matter?
Hospitality tests whether the structure can accept a new coordinate or subordinate class without changing unrelated meanings. Add plausible future values and boundary cases. If every addition needs an exception or wholesale renumbering, revise the characteristic, value set or relationships.
Assessment move
Take any navigation list and write the universe plus the characteristic behind each label. Split mixed dimensions, distinguish arrays from chains and add a difficult new item. Rehearse the repair sequence rather than memorising only the final tree. For every proposed class, write an inclusion rule, an exclusion rule and one borderline object.
Mark whether the relationship is hierarchical, coordinate, associative or faceted before drawing a diagram. If sibling labels answer different questions, extract their characteristics and rebuild the model as separate dimensions. Test exhaustivity by adding a plausible item that was not present during design, then decide whether the schema needs another value, a repeatable value or a clarified universe.
Test mutual exclusivity only where the use requires it; do not force a single placement when multiple access routes are meaningful. Compare the conceptual schema with its navigation presentation so that a clean menu is not mistaken for a sound model. Keep notes on every exception and decide whether it reveals a bad rule, a genuine overlap or a data-quality problem.
Finish by explaining how the revised structure supports retrieval and what evidence would challenge the chosen boundaries.
