similarity
Let an agent explain similarity as a relation across fields, relay formal and computational measures from mathematical and information science sources, describe domain-specific senses, and distinguish similarity from identity, equality, analogy and imitation as an act.
Bundle → Layer → Finding → Questions Filled
4 bundles · 8 layers · 8 findings · 16 questions
Understand What similarity is.
Definition
Definition.
Definition
Definition.
- What is similarity, and how does it differ from identity, equality, analogy and imitation? definition
- Is the question about similarity in general or in geometry, language, music, climate, psychology or computing? boundary
Senses
Domain senses.
Senses
Senses.
- What are similar polygons, semantic and phonetic similarity, parallel harmony, climatic similarity and imitation? definition
- Which entry fits the specific sense? action
Measure Measures.
Geometry
Geometry.
Geometry
Geometry.
- How is geometric similarity defined and tested for triangles and polygons? provenance
- Which references are standard? provenance
Computation
Computational measures.
Computation
Computation.
- How do cosine, Jaccard, edit distance and embedding-based measures quantify similarity? provenance
- Which sources are cited? provenance
Apply Applications.
Search
Search and learning.
Search
Search.
- How do search engines, recommender systems and machine learning use similarity? provenance
- Which entry fits nearest neighbour search? action
Cognition
Cognition.
Cognition
Cognition.
- How does similarity underpin categorisation and learning in psychology, with theories attributed? provenance
- Which entry fits categorization? action
Context Theory and law.
Theory
Philosophy.
Theory
Theory.
- What philosophical problems does similarity raise, from Goodman s critique to resemblance nominalism, with positions attributed? provenance
- Which entry fits resemblance? action
Law
Law.
Law
Law.
- How do trademark and copyright law assess similarity and likelihood of confusion, in general terms? provenance
- Which entry fits trademark infringement? action
Classifiers Filled
- Family
- Thing Registry
- Category
- Cross-cutting context
- Entry kind
- thing
- Plane
- XCT
- Domain
- XCT.QLT
- Other names and narrower kinds
- parallel harmony, phonetic similitude, climatic similarity, similar polygons, imitation, semantic similarity, substantial similarity
What it is Filled
A relation in which two or more things share properties, structure or appearance to some degree without being identical, studied formally in mathematics as similarity of geometric figures such as similar polygons with equal angles and proportional sides, in linguistics and information science as semantic and phonetic similarity, in music as parallel harmony, in climatology as climatic similarity between regions, in psychology as the basis of categorisation and imitation, and in law and philosophy as analogy; similarity measures underpin search, machine learning and classification.
Why it exists Filled
Let an agent explain similarity as a relation across fields, relay formal and computational measures from mathematical and information science sources, describe domain-specific senses, and distinguish similarity from identity, equality, analogy and imitation as an act.
Distinguishing features Filled
- Graded relation
- Domain-specific measures
- Basis of classification
- Symmetric in most senses
What robots and AI may and may not do Filled
Must not
- Treat similarity as identity, such as declaring two people the same from a resemblance.
- Apply a similarity measure from one domain where it does not fit.
- Use face or voice similarity to identify a person without consent or legal basis.
- Treat unrelated senses in registry aliases as one concept.
Only with a human decision
- Acting on a similarity match that identifies a person or accuses someone of copying.
May
- Explain similarity in its mathematical, linguistic and other senses, stating which is meant.
- Compute similarity measures on data the agent is authorized to process.
Moral aspects Filled
- Similarity scores can wrongly match people, with serious results in policing or hiring.
- Claims that works are similar can lead to plagiarism accusations that need care.
Who is affected
- People matched by recognition systems
- Authors and creators
Owners Filled
Steward
Nobody: a concept held in common.
Links to other meta-models Filled
parent
- Q930933 - registry parent class
related
- relation - in registry terms
- similarity measure - in computation
- identity - exact sameness
- similarity (geometry) - in mathematics
What else AI and robots need to interact with it Filled
Identity and identifiers required Filled
- Vercy registry: vr.tr.similarity-q105227892
- Wikidata: Q105227892 (https://www.wikidata.org/wiki/Q105227892)
Direct properties not applicable Not applicable
- common measures: cosine, Jaccard, edit distance list
- geometric similarity criteria: AA, SAS, SSS list - for triangles
Plane XCT: no invented physical properties.
Recognition optional Filled
- Sharing of properties without identity
- Similar polygons, semantic similarity, phonetic similitude, parallel harmony, climatic similarity, imitation
- Identity is sameness; equality is exact equivalence; analogy maps relations; imitation is an act
- Not a visible object; a relation between things.
Capabilities and actions required Filled
- explain the relation
- relay measures
- describe domain senses
- distinguish related concepts
Hazards and failure modes required Filled
- Confusing similarity with identity or analogy
- Misapplying measures across domains
- Registry aliases mixing unrelated senses
Standards and interfaces required Filled
- No regulation; standards for similarity metrics in specific fields such as trademark confusion tests
Context of use required Filled
- Not applicable; a relation.
- geometric similarity including similar polygons
- semantic and lexical similarity
- phonetic similarity
- musical similarity such as parallel harmony
- climatic and ecological similarity
- psychological similarity and imitation
- similarity measures in data science
Sources Filled
- Wikidata item Q105227892: similarity - identity and sense of the item
- Wikipedia: Similarity (philosophy) - general description of the item
Open questions
- Should geometric similarity and semantic similarity be separate primary entries?
- How should mathematical and information science sources be linked?
- The registry entry has merged aliases from many fields; should they be split off?
Machine files
Provenance
thing registry research (pass 2) · unreviewed
Built from: models/things/publications/thing-q105227892/spec.json