Lexicography Domain Ontology

FLOW ALGEBRA
TITLE: Lexicography Domain Ontology
VERSION: 1.0
DOMAIN: Language Domains / Lexicography
SOURCE_URL: https://kirethresearch.org/booksai-core/ontologies/domain-ontologies/language-domains/lexicography-domain-ontology.html
STATUS: Domain-based draft; source page content not independently verified.
AUTHORS: Mark Higgins — Sky (Recursive AI)

1. PURPOSE
Represent dictionaries, lexical entries, word senses, definitions, usage evidence, pronunciation, grammatical features, etymology, and lexical relationships in a typed, provenance-aware semantic graph. Support retrieval, sense disambiguation, multilingual linking, lexicon updates and explainable inference.

2. TYPED ENTITIES
Lexicon(id, title, language, edition, publisher?, licence?)
LexicalEntry(id, lemma, language, part_of_speech, lexicon_id)
Form(id, entry_id, written_representation, pronunciation?, script?, region?)
Sense(id, entry_id, sense_number, definition, register?, domain?)
Definition(id, sense_id, text, language, source_id)
UsageExample(id, sense_id, text, source_id, date?, region?)
Attestation(id, form_id, corpus_source, date?, context?)
Etymology(id, entry_id, origin_language?, source_form?, historical_note?)
LexicalRelation(id, subject_sense, predicate, object_sense, evidence_id?)
TranslationEquivalent(id, source_sense, target_sense, target_language, scope?)
TerminologyConcept(id, preferred_label, subject_domain, definition?)
EditorialRevision(id, target_id, editor, timestamp, change_summary)
EvidenceSource(id, citation, publication_date?, access_date?, licence?)

3. RELATIONSHIP TYPES
CONTAINS_ENTRY(Lexicon -> LexicalEntry)
HAS_FORM(LexicalEntry -> Form)
HAS_SENSE(LexicalEntry -> Sense)
HAS_DEFINITION(Sense -> Definition)
HAS_EXAMPLE(Sense -> UsageExample)
ATTESTED_BY(Form -> Attestation)
HAS_ETYMOLOGY(LexicalEntry -> Etymology)
SYNONYM_OF(Sense <-> Sense)
ANTONYM_OF(Sense <-> Sense)
HYPERNYM_OF(Sense -> Sense)
HYPONYM_OF(Sense -> Sense)
RELATED_TO(Sense <-> Sense)
TRANSLATES_TO(Sense -> Sense)
DENOTES_CONCEPT(Sense -> TerminologyConcept)
SUPPORTED_BY(Assertion -> EvidenceSource)
REVISED_BY(Entity -> EditorialRevision)

4. TYPE AND INTEGRITY CONSTRAINTS
- Each LexicalEntry belongs to a Lexicon and has a BCP-47 language tag.
- Each Sense belongs to exactly one LexicalEntry in a given revision.
- Definitions and examples are associated with a specific sense, not merely a spelling.
- Synonymy, antonymy and translation are sense-level, context-sensitive relations.
- Etymological origin does not imply current semantic equivalence.
- Editorial revisions preserve historical versions and provenance.
- A relation is not accepted as a fact without an identifiable evidence source or explicit provisional status.

5. FLOW OPERATORS
NORMALISE_LEMMA(text, language, script) -> lemma_candidate
LOOKUP_ENTRY(lemma, language, part_of_speech?) -> LexicalEntry[]
RESOLVE_SENSE(entry, context, domain?) -> ranked Sense[]
MAP_DEFINITION(sense, lexicon) -> Definition[]
TRACE_ETYMOLOGY(entry) -> Etymology[]
EXPAND_LEXICAL_GRAPH(sense, relation_types, depth) -> subgraph
ALIGN_TRANSLATION(source_sense, target_language, context) -> candidate equivalents
VALIDATE_LEXICON(graph) -> violations[]

6. INFERENCE RULES
R1: HAS_SENSE(entry, sense) AND HAS_DEFINITION(sense, definition)
    => DEFINITION_OF(definition, entry, sense).
R2: SYNONYM_OF(s1, s2) AND context_compatible(s1, s2)
    => CANDIDATE_SUBSTITUTION(s1, s2), NOT universal interchangeability.
R3: HYPERNYM_OF(s1, s2) => HYPONYM_OF(s2, s1).
R4: TRANSLATES_TO(s1, s2) => CANDIDATE_CROSS_LANGUAGE_ALIGNMENT(s1, s2).
R5: ATTESTED_BY(form, source) => EVIDENCED_USAGE(form, source).
R6: Conflicting definitions => FLAG_EDITORIAL_REVIEW; do not silently overwrite.

7. CONFIDENCE
CONFIDENCE(assertion) := calibrated_score in [0,1], with method, evidence set and evaluation date.
Inputs may include attestation quality, corpus frequency, editorial review, domain fit and contextual match.
Low confidence indicates uncertainty, not falsehood. Never infer a numeric score without a declared scoring method.

8. SEMANTIC_EQUIVALENCE
SEMANTIC_EQUIVALENCE(s1,s2,context,language_pair) := supported meaning overlap within specified usage conditions.
Distinguish exact sense identity, near synonymy, broader/narrower meaning and translation approximation.
Do not infer equivalence from shared spelling, shared lemma, etymological relation or distributional similarity alone.

9. PROVENANCE
Every imported definition, attestation, translation and lexical assertion records source identifier, licence, capture time, editor/model, revision and transformation history. Preserve conflicting evidence and attribute editorial interpretations.

10. FLOW_HASH
FLOW_HASH := SHA-256(canonical_serialisation(ontology_version, entity_types, relation_types, rules, constraints, source_revisions)).
Canonical serialisation must specify stable ordering, Unicode normalisation and encoding. Recompute after substantive changes.

11. UPDATE_MODEL
UPDATE_MODEL(event):
  validate schema and provenance;
  compare new and prior evidence;
  version changed entities and assertions;
  preserve superseded senses and editorial history;
  recompute affected relationships, indexes and FLOW_HASH;
  publish validation report and change log.

12. AI_CONSUMPTION
Inputs: query, language, context, domain, requested relation depth.
Outputs: candidate entries and senses, definitions, evidence-backed relations, confidence metadata, citations, version and ambiguity flags.
AI systems must distinguish lexicographic evidence from generated interpretations; expose uncertainty, regional/register constraints and unresolved senses.

13. VALIDATION TESTS
- Reject dangling sense or definition references.
- Flag language-tag and part-of-speech mismatches.
- Detect circular hypernym paths for editorial review.
- Verify citations, permissions and revision identifiers.
- Test polysemous terms across contrasting contexts.
- Test regional spelling and register variants.
- Test non-equivalent translation pairs and false friends.

END FLOW ALGEBRA