classification
Let an agent explain classification as a process and its principles, describe classification schemes across fields, support designing and applying classifications, and relate classification to machine learning.
Bundle → Layer → Finding → Questions Filled
4 bundles · 8 layers · 8 findings · 16 questions
Understand What classification is.
Principles
Principles.
Principles
Principles.
- What makes a good classification, including mutually exclusive and exhaustive classes and clear criteria? definition
- Which field sense is meant here? boundary
Schemes
Schemes across fields.
Schemes
Schemes.
- How do biological, library, statistical, seismic, historical and other classification schemes differ? provenance
- Which entry fits a specific scheme? action
Design Designing and applying.
Design
Designing a classification.
Design
Design.
- How can a classification scheme be designed for this purpose, and how are edge cases handled? action
- Which entry fits taxonomy design? action
Apply
Classifying items.
Apply
Application.
- How should this item be classified under this scheme, and how is consistency ensured? action
- Which references decide? provenance
Computing Automated classification.
ML
Machine learning classifiers.
ML
Machine learning.
- How do text, image and other classifiers work, and how are they evaluated? provenance
- Which entry fits machine learning? action
Bias
Bias and fairness.
Bias
Bias.
- What biases arise in classification systems and classifiers, and what do researchers recommend? provenance
- Which findings are contested? boundary
Learn History and teaching.
History
History of classification.
History
History.
- How did classification develop from Aristotle and Linnaeus to modern knowledge organisation, including traditions such as hadith classification? provenance
- Which references are standard? provenance
Teach
Teaching.
Teach
Teaching.
- How can classification be taught across subjects? action
- Which misconceptions arise? provenance
Classifiers Filled
- Family
- Thing Registry
- Category
- Cross-cutting context
- Entry kind
- thing
- Plane
- XCT
- Domain
- XCT.QLT
- Other names and narrower kinds
- age periodization, musannaf, classification of cells, text classification, seismic classification, music competition class, eight-constitutions, RNA classification, security level, quasi-genre, position classification, functional classification
What it is Filled
The process of assigning things to classes or categories according to shared characteristics, and the resulting schemes, used in science for organisms, cells and rocks, in information handling for texts, documents and images, in history for periodisation, in seismology, in music competitions and in traditional scholarship such as hadith collections; classification underpins knowledge organisation, statistics and machine learning and is evaluated for consistency and usefulness.
Why it exists Filled
Let an agent explain classification as a process and its principles, describe classification schemes across fields, support designing and applying classifications, and relate classification to machine learning.
Distinguishing features Filled
- Criteria-based
- Hierarchical or flat
- Cross-field
- Evaluable
What robots and AI may and may not do Filled
Must not
- Classify people in ways that reproduce bias.
- Hide misclassification rates.
- Use outdated schemes without saying so.
- Apply classifications to people's sensitive traits without consent.
Only with a human decision
- Classifications that decide people's access to services, credit or rights.
May
- Build and apply classification schemes with clear criteria.
- Report the scheme and version used for each classification.
- Flag ambiguous or borderline cases.
Moral aspects Filled
- Classifications of people have been used to discriminate.
- Misclassification can hide risks or deny help.
Who is affected
- People classified
- Users of the scheme
Owners Filled
Steward
Nobody owns the process; the owner of each scheme answers for it.
Links to other meta-models Filled
parent
- Q29182544 - registry parent class
- Q3249551 - registry parent class
- Q99194088 - registry parent class
related
- process - category
- information handling - category
- equivalence relation - classes as equivalence classes
- abstraction - forming categories
What else AI and robots need to interact with it Filled
Identity and identifiers required Filled
- Vercy registry: vr.tr.classification
- Wikidata: Q13582682 (https://www.wikidata.org/wiki/Q13582682)
Direct properties not applicable Not applicable
Not applicable
Plane XCT: no invented physical properties.
Recognition optional Filled
- Assignment to classes by characteristics
- Schemes and criteria
- Clustering finds groups without predefined classes; ranking orders items
- Not physical; schemes, taxonomies and labelled sets.
Capabilities and actions required Filled
- explain principles
- describe schemes across fields
- support designing classifications
- relate to machine learning
Hazards and failure modes required Filled
- Ambiguous or overlapping classes
- Bias in classification systems
- Misclassification
Standards and interfaces required Filled
- Standard classification systems such as library and statistical classifications
- Taxonomic codes of nomenclature
- Evaluation standards for classifiers
Context of use required Filled
- Assigning things to categories.
- biological and scientific classification
- library and document classification
- text and image classification in computing
- periodisation and historical classification
- classification in competitions and administration
Sources Filled
- Wikidata item Q13582682: classification - identity and sense of the item
- Wikipedia: Classification - general description of the item
Open questions
- Should each classification system be a separate entry?
- How should standards be linked?
- How should machine learning resources be linked?
Machine files
Provenance
thing registry research (pass 2) · unreviewed
Built from: models/things/publications/thing-q13582682/spec.json