Canonical Concepts

CogniSwitch concept registry

A draft registry of canonical concept IDs used by CogniSwitch structured data. Pages link to these IDs instead of minting local concepts for the same idea.

41registered concepts
8coined terms
0reconciliation TODOs

Registry status

This page publishes the current registry draft. External reconciliation fields are still being completed, so treat the registry as canonical for CogniSwitch IDs, not as fully reconciled with Wikidata or healthcare domain ontologies.

View cs: vocabulary

Proposal queue

New concepts enter through the append-only proposal queue before a Registrar promotes them into canonical concept IDs. The current local queue has 0 proposed concepts awaiting review.

AI Guardrails

COINED for this sense — the constraints applied to an AI system's inputs and outputs so it cannot act outside policy. Deliberately NO sameAs: Wikidata has no entity for this meaning, and the nearest matches (Q3095365, Q1188866) are highway barriers. Registered 2026-10-24 after a sweep found 120 mentions across 10 pages, including a dedicated /guardrails route, while the concept was absent from the registry. prefLabel is "AI Guardrails" to keep the entity away from the traffic sense.

https://cogniswitch.ai/concepts#ai-guardrails

coined
guardrailguardrailsAI guardrailcontent guardrails

COINED for this sense — the constraints applied to an AI system's inputs and outputs so it cannot act outside policy. Deliberately NO sameAs: Wikidata has no entity for this meaning, and the nearest matches (Q3095365, Q1188866) are highway barriers. Registered 2026-10-24 after a sweep found 120 mentions across 10 pages, including a dedicated /guardrails route, while the concept was absent from the registry. prefLabel is "AI Guardrails" to keep the entity away from the traffic sense.

Appears on
/deterministic-vs-probabilistic-guardrails/evals-vs-guardrails-vs-governance/guardrails/galileo-and-cogniswitch/best-llm-eval-observability-tools/arize-and-cogniswitch/braintrust-and-cogniswitch/cekura-and-cogniswitch

Context Graph

COINED by CogniSwitch; governance-layer knowledge graph where the LLM is removed from retrieval and reasoning. Anchored to knowledge-graph lineage; no external equivalent.

https://cogniswitch.ai/concepts#context-graph

coined
context graphs

COINED by CogniSwitch; governance-layer knowledge graph where the LLM is removed from retrieval and reasoning. Anchored to knowledge-graph lineage; no external equivalent.

Appears on
/blog/context-graphs-eli5/clarity/context-graphs-clarity/guides/graph-database-vs-knowledge-graph

Data Provenance

Canonical CogniSwitch concept registry entry for Data Provenance.

https://cogniswitch.ai/concepts#provenance

provenanceaudit trailaudit trailsprovenance tracking
sameAs
https://en.wikipedia.org/wiki/Provenance#Data_provenance
Appears on
/guides/graph-database-vs-knowledge-graph/governance-blind-spot/blog/context-as-first-class-citizen/telehealth/telehealth/visit-level-quality-assurance/telehealth/provider-performance-protection/telehealth/clinical-stewardship-monitoring/telehealth/vp-clinical-quality

DataOps

Canonical CogniSwitch concept registry entry for DataOps.

https://cogniswitch.ai/concepts#dataops

sameAs
https://en.wikipedia.org/wiki/DataOps
Wikidata
Q30591394

Deterministic Reasoning

Canonical CogniSwitch concept registry entry for Deterministic Reasoning.

https://cogniswitch.ai/concepts#deterministic-reasoning

deterministic reasoningdeterministic structure
Appears on
/guides/graph-database-vs-knowledge-graph/neuro-symbolic/governance-diagnostic/telehealth/telehealth/visit-level-quality-assurance/telehealth/provider-performance-protection/telehealth/clinical-stewardship-monitoring/telehealth/patient-activation-intelligence/telehealth/population-risk-intelligence/telehealth/cmo/telehealth/vp-clinical-quality/telehealth/vp-population-health

Deterministic Retrieval

COINED by CogniSwitch; retrieval whose result is a function of the graph rather than of embedding proximity, so the same query returns the same facts. Split from deterministic-reasoning 2026-10-24 (retrieval = input side, reasoning/verification = output side); the altLabel was removed there in the same pass. No external equivalent.

https://cogniswitch.ai/concepts#deterministic-retrieval

coined
deterministic retrievalgraph-based retrievalpresence-based retrieval

COINED by CogniSwitch; retrieval whose result is a function of the graph rather than of embedding proximity, so the same query returns the same facts. Split from deterministic-reasoning 2026-10-24 (retrieval = input side, reasoning/verification = output side); the altLabel was removed there in the same pass. No external equivalent.

Appears on
/deterministic-retrieval/deterministic-retrieval/token-efficiency/deterministic-retrieval/latency

DevOps

Canonical CogniSwitch concept registry entry for DevOps.

https://cogniswitch.ai/concepts#devops

sameAs
https://en.wikipedia.org/wiki/DevOps
Wikidata
Q3025536

Explainability (artificial intelligence)

Canonical CogniSwitch concept registry entry for Explainability (artificial intelligence).

https://cogniswitch.ai/concepts#explainability

explainabilityexplainable AIXAI
sameAs
https://en.wikipedia.org/wiki/Explainable_artificial_intelligence
Wikidata
Q40890078
Appears on
/frameworks/explainability-trap/neuro-symbolic/verifiable-ai

Grounding

COINED by CogniSwitch for this specific sense: attaching a mined fact to a defined domain concept, which is what makes the fact checkable rather than merely well-formed. Deliberately NO sameAs — the nearest external entity, the symbol grounding problem (Q246989), is a cognitive-science problem about how symbols acquire meaning, and is a different concept. Adjacency is recorded here rather than asserted as equivalence. General framing credited to Casey Hart in the Vocabulary article's Recommended Reading.

https://cogniswitch.ai/concepts#grounding

coined
groundedgrounded in realityungrounded fact

COINED by CogniSwitch for this specific sense: attaching a mined fact to a defined domain concept, which is what makes the fact checkable rather than merely well-formed. Deliberately NO sameAs — the nearest external entity, the symbol grounding problem (Q246989), is a cognitive-science problem about how symbols acquire meaning, and is a different concept. Adjacency is recorded here rather than asserted as equivalence. General framing credited to Casey Hart in the Vocabulary article's Recommended Reading.

Appears on
/enterprise-ontology/vocabulary

Hallucination (artificial intelligence)

Canonical CogniSwitch concept registry entry for Hallucination (artificial intelligence).

https://cogniswitch.ai/concepts#hallucination

hallucinationhallucinationshallucinate
sameAs
https://en.wikipedia.org/wiki/Hallucination_(artificial_intelligence)
Wikidata
Q116197048
Appears on
/neuro-symbolic/governance-blind-spot/guides/graph-database-vs-knowledge-graph

Hangs as a Fact

COINED by CogniSwitch, entirely our own. Names the failure mode of a well-formed, syntactically valid fact with no defined domain concept to attach to — the exact shape of a hallucination. No external equivalent.

https://cogniswitch.ai/concepts#hangs-as-a-fact

coined
hangs as a facthomeless fact

COINED by CogniSwitch, entirely our own. Names the failure mode of a well-formed, syntactically valid fact with no defined domain concept to attach to — the exact shape of a hallucination. No external equivalent.

Broader
Grounding
Appears on
/enterprise-ontology/vocabulary

Healthcare AI

Canonical CogniSwitch concept registry entry for Healthcare AI.

https://cogniswitch.ai/concepts#healthcare-ai

healthcare AIClinical AIhealth AI
sameAs
https://en.wikipedia.org/wiki/Artificial_intelligence_in_healthcare
Wikidata
Q28324961
Appears on
/clarity/healthcare/guides/graph-database-vs-knowledge-graph/governance-blind-spot/telehealth/telehealth/visit-level-quality-assurance/telehealth/provider-performance-protection/telehealth/clinical-stewardship-monitoring/telehealth/patient-activation-intelligence/telehealth/population-risk-intelligence/telehealth/cmo/telehealth/vp-clinical-quality/telehealth/vp-population-health

LLM Evaluation and Observability Tooling

Category hub for the LLM eval/observability tooling space (Arize, Galileo, Braintrust, Cekura, et al.) that CogniSwitch is positioned against. Groups the competitor-comparison and eval-category pages so they interlink as a cluster.

https://cogniswitch.ai/concepts#llm-eval-observability

operational definition
LLM evaluationLLM evalLLM evalsLLM observabilityLLM monitoringeval observability

Category hub for the LLM eval/observability tooling space (Arize, Galileo, Braintrust, Cekura, et al.) that CogniSwitch is positioned against. Groups the competitor-comparison and eval-category pages so they interlink as a cluster.

Appears on
/best-llm-eval-observability-tools/arize-and-cogniswitch/braintrust-and-cogniswitch/cekura-and-cogniswitch/galileo-and-cogniswitch/evals-vs-guardrails-vs-governance/evals-vs-audit/deterministic-vs-probabilistic-guardrails/eval-criteria-workbench/eval-criteria-workbench/criteria-guide

LLM-as-a-Judge

CLUSTER-SPEC.md names this the canonical critique of the eval/guardrail cluster: the argument every other page in that cluster links to rather than re-deriving. 61 mentions across 8 pages, 24 on its own guide route.

https://cogniswitch.ai/concepts#llm-as-a-judge

operational definition
LLM as a judgeLLM-as-judgeLLM-based evaluationmodel-based evaluationmodel-graded evaluation

CLUSTER-SPEC.md names this the canonical critique of the eval/guardrail cluster: the argument every other page in that cluster links to rather than re-deriving. 61 mentions across 8 pages, 24 on its own guide route.

sameAs
https://en.wikipedia.org/wiki/LLM-as-a-Judge
Wikidata
Q134468621
Appears on
/guides/llm-as-a-judge-vs-deterministic-verification/deterministic-vs-probabilistic-guardrails/braintrust-and-cogniswitch/arize-and-cogniswitch/best-llm-eval-observability-tools/galileo-and-cogniswitch/evals-vs-guardrails-vs-governance

MLOps

Canonical CogniSwitch concept registry entry for MLOps.

https://cogniswitch.ai/concepts#mlops

sameAs
https://en.wikipedia.org/wiki/MLOps
Wikidata
Q60753505

Neuro-Symbolic AI

Canonical CogniSwitch concept registry entry for Neuro-Symbolic AI.

https://cogniswitch.ai/concepts#neuro-symbolic-ai

neurosymbolic AIneuro symbolicNeuro-symbolic
sameAs
https://en.wikipedia.org/wiki/Neuro-symbolic_AI
Wikidata
Q113512183
Appears on
/neuro-symbolic/blog/neuro-symbolic-ai/blog/neuro-symbolic-eli5

Phantom Human-in-the-Loop

COINED by CogniSwitch. Anti-pattern: human oversight is nominally present in an AI pipeline but structurally impossible at production throughput, opacity, or attention scale. The human exists on paper; the review does not happen in practice.

https://cogniswitch.ai/concepts#phantom-human-in-the-loop

coined
Phantom HITLphantom human in the loop

COINED by CogniSwitch. Anti-pattern: human oversight is nominally present in an AI pipeline but structurally impossible at production throughput, opacity, or attention scale. The human exists on paper; the review does not happen in practice.

Appears on
/blog/phantom-human-in-the-loop/human-oversight/blog/evals-are-not-audits

Property Graph

Disambiguated 2026-10-24: Q109038369, the data model used by graph-oriented databases — the sense the ContextOps guests use when contrasting property graphs against RDF. NOT Q113301322, a narrower formal definition with no Wikipedia article.

https://cogniswitch.ai/concepts#property-graph

operational definition
property graphproperty graphslabeled property graph

Disambiguated 2026-10-24: Q109038369, the data model used by graph-oriented databases — the sense the ContextOps guests use when contrasting property graphs against RDF. NOT Q113301322, a narrower formal definition with no Wikipedia article.

sameAs
https://en.wikipedia.org/wiki/Property_graph
Wikidata
Q109038369
Appears on
/contextops/stephane-fellah-ontologies-geospatial/contextops/melli-annamalai-one-tool/contextops/karthik-soman-different-kind-of-answer/contextops/himanshu-singh-your-graph-is-not-your-database/contextops/casey-hart-say-true-things

Retrieval-Augmented Generation

Canonical CogniSwitch concept registry entry for Retrieval-Augmented Generation.

https://cogniswitch.ai/concepts#retrieval-augmented-generation

RAGVector RAG
sameAs
https://en.wikipedia.org/wiki/Retrieval-augmented_generation
Wikidata
Q121362277
Appears on
/neuro-symbolic/guides/graph-database-vs-knowledge-graph

Semantic Web

Canonical CogniSwitch concept registry entry for Semantic Web.

https://cogniswitch.ai/concepts#semantic-web

semantic webSemWeb
sameAs
https://en.wikipedia.org/wiki/Semantic_Web
Wikidata
Q54837
Appears on
/contextops/stephane-fellah-ontologies-geospatial/contextops/jessica-talisman-context-lives-in-relationships/contextops/melli-annamalai-one-tool/contextops/casey-hart-say-true-things/blog/launching-kl4a-knowledge-layer-for-agents

Shapes Constraint Language

Canonical CogniSwitch concept registry entry for Shapes Constraint Language.

https://cogniswitch.ai/concepts#shacl

SHACLshapes constraint language
sameAs
https://en.wikipedia.org/wiki/SHACL
Wikidata
Q29377821
Appears on
/contextops/stephane-fellah-ontologies-geospatial/contextops/jessica-talisman-context-lives-in-relationships/contextops/elliott-risch-own-your-meaning/contextops/casey-hart-say-true-things

Simple Knowledge Organization System

The W3C vocabulary this registry is itself written in — every concept here carries skos:prefLabel, skos:Concept and skos:inScheme. Registered 2026-10-24; it had been the registry's own modelling language while absent from it.

https://cogniswitch.ai/concepts#skos

operational definition
SKOSSimple Knowledge Organisation System

The W3C vocabulary this registry is itself written in — every concept here carries skos:prefLabel, skos:Concept and skos:inScheme. Registered 2026-10-24; it had been the registry's own modelling language while absent from it.

sameAs
https://en.wikipedia.org/wiki/Simple_Knowledge_Organization_System
Wikidata
Q2288360
Appears on
/contextops/jessica-talisman-context-lives-in-relationships/contextops/stephane-fellah-ontologies-geospatial/blog/how-to-operationalize-a-context-graph

SPARQL

Canonical CogniSwitch concept registry entry for SPARQL.

https://cogniswitch.ai/concepts#sparql

SPARQLSPARQL Protocol and RDF Query Language
sameAs
https://en.wikipedia.org/wiki/SPARQL
Wikidata
Q54871
Appears on
/contextops/stephane-fellah-ontologies-geospatial/contextops/casey-hart-say-true-things/contextops/melli-annamalai-one-tool/contextops/karthik-soman-different-kind-of-answer/contextops/jessica-talisman-context-lives-in-relationships

Taxonomy (classification)

Disambiguated 2026-10-24: Q7211, the classification of things or concepts. NOT Q8269924 (biological taxonomy, the naming of organisms).

https://cogniswitch.ai/concepts#taxonomy

operational definition
taxonomytaxonomiestaxonomic scheme

Disambiguated 2026-10-24: Q7211, the classification of things or concepts. NOT Q8269924 (biological taxonomy, the naming of organisms).

sameAs
https://en.wikipedia.org/wiki/Taxonomy
Wikidata
Q7211
Appears on
/enterprise-ontology/vocabulary

Traceability (AI output)

COINED operational definition by CogniSwitch. Traceability means mapping every AI output back to its exact source chain: document → section → concept → decision. The industry uses the term loosely; this definition is structural and auditable. 'Traceability ceiling' names the maximum level of auditability any given architecture can achieve.

https://cogniswitch.ai/concepts#traceability

coined
traceabilitytraceability ceilingtraceablefull provenance trail

COINED operational definition by CogniSwitch. Traceability means mapping every AI output back to its exact source chain: document → section → concept → decision. The industry uses the term loosely; this definition is structural and auditable. 'Traceability ceiling' names the maximum level of auditability any given architecture can achieve.

Wikidata
Q3997103
Appears on
/guides/graph-database-vs-knowledge-graph/governance-blind-spot/verifiable-ai/telehealth/telehealth/visit-level-quality-assurance/telehealth/provider-performance-protection/telehealth/clinical-stewardship-monitoring/telehealth/patient-activation-intelligence/telehealth/population-risk-intelligence/telehealth/cmo/telehealth/vp-clinical-quality/telehealth/vp-population-health