@prefix : <https://community.openlinksw.com/t/llms-and-language/6504#> .
@prefix dbr: <http://dbpedia.org/resource/> .
@prefix owl: <http://www.w3.org/2002/07/owl#> .
@prefix prov: <http://www.w3.org/ns/prov#> .
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
@prefix schema1: <http://schema.org/> .
@prefix xsd: <http://www.w3.org/2001/XMLSchema#> .

:ComparisonDimension a rdfs:Class ;
    rdfs:label "ComparisonDimension"@en ;
    schema1:description "One aspect of the implicit-versus-explicit semantics comparison: semantics, representation, identity, inference, coupling, or failure mode."@en ;
    schema1:name "ComparisonDimension"@en ;
    rdfs:comment "One aspect of the implicit-versus-explicit semantics comparison: semantics, representation, identity, inference, coupling, or failure mode."@en ;
    rdfs:isDefinedBy :llmsLanguageOntology ;
    rdfs:subClassOf schema1:CreativeWork .

:DivisionLayer a rdfs:Class ;
    rdfs:label "DivisionLayer"@en ;
    schema1:description "One layer in the essay's division of labor between probabilistic language handling and explicit, machine-computable semantics."@en ;
    schema1:name "DivisionLayer"@en ;
    rdfs:comment "One layer in the essay's division of labor between probabilistic language handling and explicit, machine-computable semantics."@en ;
    rdfs:isDefinedBy :llmsLanguageOntology ;
    rdfs:subClassOf schema1:CreativeWork .

:LanguageComponent a rdfs:Class ;
    rdfs:label "LanguageComponent"@en ;
    schema1:description "One of the four components of language in the essay's account: signs, syntax, semantics, or pragmatics — each with a counterpart realization inside a large language model."@en ;
    schema1:name "LanguageComponent"@en ;
    rdfs:comment "One of the four components of language in the essay's account: signs, syntax, semantics, or pragmatics — each with a counterpart realization inside a large language model."@en ;
    rdfs:isDefinedBy :llmsLanguageOntology ;
    rdfs:subClassOf schema1:CreativeWork .

:llmRealization a rdf:Property ;
    rdfs:label "llmRealization"@en ;
    schema1:description "What a language component becomes inside a large language model: tokens and numerical representations, learned structural regularities, learned contextual associations, or runtime context."@en ;
    schema1:domainIncludes :LanguageComponent ;
    schema1:name "llmRealization"@en ;
    schema1:rangeIncludes schema1:Text ;
    rdfs:comment "What a language component becomes inside a large language model: tokens and numerical representations, learned structural regularities, learned contextual associations, or runtime context."@en ;
    rdfs:domain :LanguageComponent ;
    rdfs:isDefinedBy :llmsLanguageOntology ;
    rdfs:range xsd:string .

:providesRole a rdf:Property ;
    rdfs:label "providesRole"@en ;
    schema1:description "The role a language component provides in the system of language: denotation, structure, meaning, or situation."@en ;
    schema1:domainIncludes :LanguageComponent ;
    schema1:name "providesRole"@en ;
    schema1:rangeIncludes schema1:Text ;
    rdfs:comment "The role a language component provides in the system of language: denotation, structure, meaning, or situation."@en ;
    rdfs:domain :LanguageComponent ;
    rdfs:isDefinedBy :llmsLanguageOntology ;
    rdfs:range xsd:string .

dbr:OpenLink_Software a schema1:Organization ;
    schema1:description "Publisher of the source article."@en ;
    schema1:name "OpenLink Software"@en ;
    schema1:url <https://www.openlinksw.com> .

:ans1 a schema1:Answer ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:text "The author's coinage for a large language model, short for Language Emulation Machine. The name is the argument: the model does not possess language the way a speaker does — it emulates the systematic use of signs, syntax, semantics, and pragmatics, reproducing their outward behavior from learned numerical patterns."@en .

:ans10 a schema1:Answer ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:text "Five layers, loosely coupled: the LLM handles probabilistic natural-language work; RDF, Linked Data, and ontologies supply explicit machine-computable semantics; logic supplies deterministic inference; the agent harness supplies controlled action and skill/tool selection; data spaces — databases, knowledge bases, filesystems, APIs — supply ground truth."@en .

:ans11 a schema1:Answer ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:text "Agents act, and acting on implicit semantics alone means acting on approximations. Grounding the agent's entities and relationships in explicit, identified, declared semantics gives it something deterministic to stand on: the fluent layer interprets intent, the precise layer constrains what may be concluded and done."@en .

:ans12 a schema1:Answer ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:text "Not smarter models — the point is that raw capability was never the gap. The missing piece was the explicit semantic layer that lets machine intelligence be checked, grounded, and composed: identified entities, declared relationships, and logic. LLMs don't make explicit semantics obsolete; they make explicit semantics easier to use."@en .

:ans2 a schema1:Answer ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:text "Signs (the carriers of meaning), syntax (the combinatorial structure), semantics (what expressions mean), and pragmatics (how situation fixes meaning). Each has a counterpart inside the model: tokens and numerical representations, learned structural regularities, learned contextual associations, and runtime context."@en .

:ans3 a schema1:Answer ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:text "That meaning is fixed by situation, not by the sign alone. The word “bank” denotes a financial institution in one context and the edge of a river in another. Pragmatics — the runtime context — selects the reading. The model does the same work with its context window that a speaker does with a situation."@en .

:ans4 a schema1:Answer ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:text "Signs become tokens and numerical representations; syntax becomes learned structural regularities absorbed from training text; semantics becomes learned contextual associations between those representations; pragmatics becomes the runtime context — the prompt, the conversation, the task. Nothing is declared; everything is encoded."@en .

:ans5 a schema1:Answer ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:text "Nothing is false about it, but it undersells what the prediction machinery achieves. Next-token prediction is the training objective; language emulation — reproducing the systematic behavior of signs, syntax, semantics, and pragmatics — is what that objective produces at scale. “Langulator” names the product, not the mechanism."@en .

:ans6 a schema1:Answer ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:text "Meaning that is encoded but not declared. In an LLM, what a token “means” exists only as patterns across numerical representations — recoverable by running the model, unavailable to inspection, and closed to deterministic inference. Fluent, approximate, and fundamentally probabilistic."@en .

:ans7 a schema1:Answer ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:text "Meaning that is identified and declared: entities named by URIs, relationships stated as RDF triples, vocabularies shared through ontologies. Explicit semantics are machine-computable — a logic engine can apply rules to them and derive conclusions that follow necessarily."@en .

:ans8 a schema1:Answer ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:text "The difference between the two kinds of semantics. An LLM associates “Paris” with “France” through learned contextual association — strong, fluent, and implicit. Explicit semantics instead declares the relationship (Paris is the capital of France) as an identified, declared, machine-computable fact that logic can build on."@en .

:ans9 a schema1:Answer ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:text "The central image: language describes the world the way a map describes terrain. A map is useful precisely because it leaves things out — and dangerous when mistaken for the ground itself. Confusing an LLM's fluent map for the territory is the characteristic failure mode."@en .

:dimCoupling a :ComparisonDimension ;
    schema1:description "How the fuzzy and the precise relate: fused inside the model versus loose coupling between separate layers."@en ;
    schema1:isPartOf :article ;
    schema1:name "Coupling"@en .

:dimFailureMode a :ComparisonDimension ;
    schema1:description "What goes wrong characteristically: fluent output mistaken for ground truth versus statements machines can check, ground, and compose."@en ;
    schema1:isPartOf :article ;
    schema1:name "Failure mode"@en .

:dimIdentity a :ComparisonDimension ;
    schema1:description "Whether a thing stays the same thing: statistical proximity versus stable identity through IRIs."@en ;
    schema1:isPartOf :article ;
    schema1:name "Identity"@en .

:dimInference a :ComparisonDimension ;
    schema1:description "How conclusions are reached: probabilistic generation versus deterministic logic over declared relationships."@en ;
    schema1:isPartOf :article ;
    schema1:name "Inference"@en .

:dimRepresentation a :ComparisonDimension ;
    schema1:description "What stands for a thing: tokens and embeddings versus URIs and triples."@en ;
    schema1:isPartOf :article ;
    schema1:name "Representation"@en .

:dimSemantics a :ComparisonDimension ;
    schema1:description "How meaning is carried: numerically encoded in model weights versus identified by URIs and declared as RDF."@en ;
    schema1:isPartOf :article ;
    schema1:name "Semantics"@en .

:kgExplorerSection a schema1:CreativeWork ;
    schema1:isPartOf :article ;
    schema1:name "Knowledge graph explorer"@en .

:recipeComponents a schema1:SoftwareSourceCode ;
    schema1:codeSampleType "full"@en ;
    schema1:description "Each LanguageComponent with the role it provides and what it becomes inside the Langulator."@en ;
    schema1:isPartOf :sparqlWorkbenchSection ;
    schema1:name "The four language components and their LLM realizations"@en ;
    schema1:programmingLanguage "SPARQL"@en ;
    schema1:text """PREFIX schema: <http://schema.org/>
PREFIX : <https://community.openlinksw.com/t/llms-and-language/6504#>

SELECT ?component ?provides ?realization
WHERE {
    GRAPH <https://linkeddata.uriburner.com/DAV/demos/daas/llms-and-language-muse.ttl> {
        ?component a :LanguageComponent ;
                   schema:name ?name ;
                   :providesRole ?provides ;
                   :llmRealization ?realization .
    }
}
ORDER BY ?name""" .

:recipeEntityTypes a schema1:SoftwareSourceCode ;
    schema1:codeSampleType "full"@en ;
    schema1:description "SAMPLE-based per-type census of the collection graph: type, one sample entity, its label, and the count — ordered by frequency."@en ;
    schema1:isPartOf :sparqlWorkbenchSection ;
    schema1:name "Entity-type summary (canonical)"@en ;
    schema1:programmingLanguage "SPARQL"@en ;
    schema1:text """PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX schema: <http://schema.org/>

SELECT
    ?type
    (SAMPLE(?s) AS ?sampleEntity)
    (SAMPLE(?label) AS ?sampleLabel)
    (COUNT(?s) AS ?entityCount)
WHERE {
    GRAPH <https://linkeddata.uriburner.com/DAV/demos/daas/llms-and-language-muse.ttl> {
        ?s rdf:type ?type .
        OPTIONAL { ?s rdfs:label ?label }
    }
}
GROUP BY ?type
ORDER BY DESC(?entityCount)""" .

:recipeFaqGlossary a schema1:SoftwareSourceCode ;
    schema1:codeSampleType "full"@en ;
    schema1:description "All twelve named questions joined to their accepted answers."@en ;
    schema1:isPartOf :sparqlWorkbenchSection ;
    schema1:name "FAQ questions with their answers"@en ;
    schema1:programmingLanguage "SPARQL"@en ;
    schema1:text """PREFIX schema: <http://schema.org/>

SELECT ?question ?answer
WHERE {
    GRAPH <https://linkeddata.uriburner.com/DAV/demos/daas/llms-and-language-muse.ttl> {
        ?q a schema:Question ;
           schema:name ?question ;
           schema:acceptedAnswer ?a .
        ?a schema:text ?answer .
    }
}
ORDER BY ?question""" .

:recipeHowto a schema1:SoftwareSourceCode ;
    schema1:codeSampleType "full"@en ;
    schema1:description "The nine HowTo steps from human intent to human-readable outcome, with positions."@en ;
    schema1:isPartOf :sparqlWorkbenchSection ;
    schema1:name "The division-of-labor pipeline in order"@en ;
    schema1:programmingLanguage "SPARQL"@en ;
    schema1:text """PREFIX schema: <http://schema.org/>

SELECT ?step ?title ?position
WHERE {
    GRAPH <https://linkeddata.uriburner.com/DAV/demos/daas/llms-and-language-muse.ttl> {
        ?step a schema:HowToStep ;
              schema:name ?title ;
              schema:position ?position .
    }
}
ORDER BY ?position""" .

:compPragmatics a :LanguageComponent ;
    schema1:description "Pragmatics is how situation fixes meaning: who is speaking, to whom, about what, right now — the difference between a financial bank and a river bank. For the model, that situation is the runtime context: the prompt, the conversation history, the task."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "Pragmatics"@en ;
    schema1:position 4 ;
    :llmRealization "Runtime context"@en ;
    :providesRole "situation — meaning in context of use"@en .

:compSemantics a :LanguageComponent ;
    schema1:description "Semantics is what signs and structures mean — reference, sense, truth conditions. In the model, meaning is not declared anywhere; it is encoded implicitly as contextual associations between numerical representations, recoverable only by running the model."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "Semantics"@en ;
    schema1:position 3 ;
    :llmRealization "Learned contextual associations"@en ;
    :providesRole "meaning — what expressions denote"@en .

:compSigns a :LanguageComponent ;
    schema1:description "Signs are the perceptible forms that carry meaning: words, marks, sounds. Inside a large language model they become tokens — sub-word units mapped to numerical vectors — the raw material everything else is computed over."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "Signs"@en ;
    schema1:position 1 ;
    :llmRealization "Tokens and numerical representations"@en ;
    :providesRole "denotation — the carriers of meaning"@en .

:compSyntax a :LanguageComponent ;
    schema1:description "Syntax is the system of rules for combining signs into well-formed expressions. The model never receives a grammar; it absorbs structural regularities statistically from the shape of training text, and reproduces them fluently."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "Syntax"@en ;
    schema1:position 2 ;
    :llmRealization "Learned structural regularities"@en ;
    :providesRole "structure — the combinatorial rules"@en .

:divDataSpaces a :DivisionLayer ;
    schema1:description "Databases, knowledge bases, filesystems, APIs. Where the ground truth lives. Documents, tables, graphs, and services — the territory the map describes, reachable through governed interfaces."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "Data spaces"@en ;
    schema1:position 5 .

:divHarness a :DivisionLayer ;
    schema1:description "Controlled action and tool selection. The scaffolding that lets an agent act: choosing skills and tools, sequencing calls, enforcing policy. It couples the fluent layer to the precise layer without confusing the two."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "Agent harness"@en ;
    schema1:position 4 .

:divLLM a :DivisionLayer ;
    schema1:description "Probabilistic natural-language handling. The model emulates language: it parses intent from prose, holds the fuzzy, contextual, ambiguous parts of a task, and renders results back into natural language. Its semantics stay implicit — that is what makes it fluent, and what makes it unreliable for precision."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "LLM — the Langulator"@en ;
    schema1:position 1 .

:divLogic a :DivisionLayer ;
    schema1:description "Deterministic inference. Rules applied to declared relationships yield conclusions that follow necessarily — no sampling, no temperature. The precise parts of a task land here."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "Logic"@en ;
    schema1:position 3 .

:divSemanticWeb a :DivisionLayer ;
    schema1:description "Explicit, machine-computable semantics. Relationships identified by URIs, declared in shared vocabularies, traversable over HTTP, and available to deterministic inference. Where the model's meaning is encoded and approximate, here it is declared and exact."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "RDF + Linked Data + Ontologies"@en ;
    schema1:position 2 .

:faq1 a schema1:Question ;
    schema1:acceptedAnswer :ans1 ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:mainEntityOfPage :faqSection ;
    schema1:name "What is a “Langulator”?"@en .

:faq10 a schema1:Question ;
    schema1:acceptedAnswer :ans10 ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:mainEntityOfPage :faqSection ;
    schema1:name "What is the proposed division of labor?"@en .

:faq11 a schema1:Question ;
    schema1:acceptedAnswer :ans11 ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:mainEntityOfPage :faqSection ;
    schema1:name "Why does this matter for AI agents?"@en .

:faq12 a schema1:Question ;
    schema1:acceptedAnswer :ans12 ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:mainEntityOfPage :faqSection ;
    schema1:name "What was the missing piece, if not intelligence?"@en .

:faq2 a schema1:Question ;
    schema1:acceptedAnswer :ans2 ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:mainEntityOfPage :faqSection ;
    schema1:name "What are the four components of language?"@en .

:faq3 a schema1:Question ;
    schema1:acceptedAnswer :ans3 ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:mainEntityOfPage :faqSection ;
    schema1:name "What does the “bank” example show?"@en .

:faq4 a schema1:Question ;
    schema1:acceptedAnswer :ans4 ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:mainEntityOfPage :faqSection ;
    schema1:name "How does an LLM handle each language component?"@en .

:faq5 a schema1:Question ;
    schema1:acceptedAnswer :ans5 ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:mainEntityOfPage :faqSection ;
    schema1:name "What is wrong with describing LLMs as next-token predictors?"@en .

:faq6 a schema1:Question ;
    schema1:acceptedAnswer :ans6 ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:mainEntityOfPage :faqSection ;
    schema1:name "What are implicit semantics?"@en .

:faq7 a schema1:Question ;
    schema1:acceptedAnswer :ans7 ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:mainEntityOfPage :faqSection ;
    schema1:name "What are explicit semantics?"@en .

:faq8 a schema1:Question ;
    schema1:acceptedAnswer :ans8 ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:mainEntityOfPage :faqSection ;
    schema1:name "What does the Paris/France example demonstrate?"@en .

:faq9 a schema1:Question ;
    schema1:acceptedAnswer :ans9 ;
    schema1:isPartOf :article,
        :faqSection ;
    schema1:mainEntityOfPage :faqSection ;
    schema1:name "What does “language is the map” mean?"@en .

:imageMapInfographic a schema1:ImageObject ;
    schema1:about :overviewSection ;
    schema1:caption "The essay's infographic: the symbiosis between large language models and a Semantic Web — implicit semantics meeting explicit semantics."@en ;
    schema1:contentUrl <https://www.openlinksw.com/data//infographics/llms-semantic-web-ontology-linkeddata-symbiosis.png> ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "LLMs, Semantic Web, ontology, and Linked Data symbiosis"@en .

:kgGeneratorSkill a schema1:SoftwareApplication,
        prov:SoftwareAgent ;
    schema1:description "Skill that generated this knowledge graph from the source article."@en ;
    schema1:isPartOf :aboutSection,
        :article ;
    schema1:name "kg-generator"@en ;
    schema1:url <https://github.com/OpenLinkSoftware/ai-agent-skills/tree/main/kg-generator> ;
    prov:actedOnBehalfOf <https://kingsley.idehen.net/DAV/home/kidehen/Public/YouID/link-in-bio-credentials-5/index.html#netid> .

:museSparkAgent a schema1:SoftwareApplication,
        prov:SoftwareAgent ;
    schema1:description "The Muse model-family agent that executed the build on the principal's behalf."@en ;
    schema1:isPartOf :aboutSection,
        :article ;
    schema1:name "Muse Spark"@en ;
    schema1:url <https://muse.ai> ;
    prov:actedOnBehalfOf <https://kingsley.idehen.net/DAV/home/kidehen/Public/YouID/link-in-bio-credentials-5/index.html#netid> .

:rdfInfographicSkill a schema1:SoftwareApplication,
        prov:SoftwareAgent ;
    schema1:description "Skill that rendered this knowledge graph as an interactive HTML infographic."@en ;
    schema1:isPartOf :aboutSection,
        :article ;
    schema1:name "rdf-infographic-skill"@en ;
    schema1:url <https://github.com/OpenLinkSoftware/ai-agent-skills/tree/main/rdf-infographic-skill> ;
    prov:actedOnBehalfOf <https://kingsley.idehen.net/DAV/home/kidehen/Public/YouID/link-in-bio-credentials-5/index.html#netid> .

:relAttention a schema1:CreativeWork ;
    schema1:description "Cited by the essay. Creator: Vaswani et al.."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "Attention Is All You Need"@en ;
    schema1:url <https://arxiv.org/abs/1706.03762> .

:relBert a schema1:CreativeWork ;
    schema1:description "Cited by the essay. Creator: Devlin et al.."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding"@en ;
    schema1:url <https://arxiv.org/abs/1810.04805> .

:relDistributional a schema1:CreativeWork ;
    schema1:description "Cited by the essay. Creator: Wikipedia."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "Distributional Hypothesis"@en ;
    schema1:url <https://en.wikipedia.org/wiki/Distributional_semantics> .

:relMeaning a schema1:CreativeWork ;
    schema1:description "Cited by the essay. Creator: Stanford Encyclopedia of Philosophy."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "Theories of Meaning"@en ;
    schema1:url <https://plato.stanford.edu/entries/meaning/> .

:relMikolov a schema1:CreativeWork ;
    schema1:description "Cited by the essay. Creator: Mikolov et al.."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "Efficient Estimation of Word Representations in Vector Space"@en ;
    schema1:url <https://arxiv.org/abs/1301.3781> .

:relPeirce a schema1:CreativeWork ;
    schema1:description "Cited by the essay. Creator: Charles Sanders Peirce."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "Theory of Signs"@en ;
    schema1:url <https://plato.stanford.edu/entries/peirce-semiotics/> .

:relRdfSemantics a schema1:CreativeWork ;
    schema1:description "Cited by the essay. Creator: W3C."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "RDF 1.1 Semantics"@en ;
    schema1:url <https://www.w3.org/TR/rdf11-mt/> .

:relShannon a schema1:CreativeWork ;
    schema1:description "Cited by the essay. Creator: Claude Shannon."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "A Mathematical Theory of Communication"@en ;
    schema1:url <https://ieeexplore.ieee.org/document/6773024> .

:step1 a schema1:HowToStep ;
    schema1:isPartOf :article,
        :howtoSection ;
    schema1:name "Start from the human"@en ;
    schema1:position 1 ;
    schema1:text "Every task begins with a person and an intent. The pipeline's job is to carry that intent through machinery without losing it — which is why the first and last steps are human."@en .

:step2 a schema1:HowToStep ;
    schema1:isPartOf :article,
        :howtoSection ;
    schema1:name "Express the intent in natural language"@en ;
    schema1:position 2 ;
    schema1:text "The human states what they want in ordinary prose. Natural language is the highest-bandwidth interface a person has; the pipeline meets them there instead of demanding a query language."@en .

:step3 a schema1:HowToStep ;
    schema1:isPartOf :article,
        :howtoSection ;
    schema1:name "Let the Langulator parse the fuzzy parts"@en ;
    schema1:position 3 ;
    schema1:text "The LLM — the Language Emulation Machine — handles everything probabilistic: disambiguating intent, filling in context, tolerating ambiguity. Its implicit semantics are a feature here, not a bug."@en .

:step4 a schema1:HowToStep ;
    schema1:isPartOf :article,
        :howtoSection ;
    schema1:name "Ground entities in RDF, Linked Data, and ontologies"@en ;
    schema1:position 4 ;
    schema1:text "Translate the parsed intent into identified things and declared relationships: URIs for entities, RDF for the relations, ontologies for shared meaning. This is where implicit becomes explicit."@en .

:step5 a schema1:HowToStep ;
    schema1:isPartOf :article,
        :howtoSection ;
    schema1:name "Operate over data spaces"@en ;
    schema1:position 5 ;
    schema1:text "Run the precise parts against databases, knowledge bases, filesystems, and APIs — the territory, not the map. Governed access decides what the agent may touch."@en .

:step6 a schema1:HowToStep ;
    schema1:isPartOf :article,
        :howtoSection ;
    schema1:name "Apply logic for deterministic results"@en ;
    schema1:position 6 ;
    schema1:text "Where the answer must follow necessarily, use inference over declared relationships — not sampling. Precision is computed, not guessed."@en .

:step7 a schema1:HowToStep ;
    schema1:isPartOf :article,
        :howtoSection ;
    schema1:name "Return through the Langulator"@en ;
    schema1:position 7 ;
    schema1:text "Hand the precise results back to the model for rendering: summaries, explanations, prose shaped for the human's situation. Fluency is the model's home turf."@en .

:step8 a schema1:HowToStep ;
    schema1:isPartOf :article,
        :howtoSection ;
    schema1:name "Deliver in natural language"@en ;
    schema1:position 8 ;
    schema1:text "The human receives the outcome as language — the same medium the intent arrived in. The machinery in the middle stays invisible."@en .

:step9 a schema1:HowToStep ;
    schema1:isPartOf :article,
        :howtoSection ;
    schema1:name "Keep the coupling loose"@en ;
    schema1:position 9 ;
    schema1:text "The standing rule: language for the fuzzy parts, a Semantic Web for the precise parts, and only loose coupling between them. Never let the probabilistic layer pretend to be the precise one."@en .

<https://community.openlinksw.com/t/llms-and-language/6504> a schema1:CreativeWork ;
    schema1:about :llmsLanguageOntology ;
    schema1:description "The OpenLink community article this collection is distilled from: language as the systematic use of signs, syntax, semantics, and pragmatics; the LLM as a Language Emulation Machine (Langulator); implicit versus explicit semantics; and the division of labor between probabilistic language handling and machine-computable meaning."@en ;
    schema1:name "LLMs and Language — community post"@en ;
    schema1:url <https://community.openlinksw.com/t/llms-and-language/6504> .

:termDataSpace a schema1:DefinedTerm ;
    schema1:description "The term for where ground truth lives: databases, knowledge bases, filesystems, and APIs — the territory the language-map describes, reached through governed interfaces."@en ;
    schema1:inDefinedTermSet :glossaryTermSet ;
    schema1:isPartOf :article,
        :glossarySection,
        :glossaryTermSet ;
    schema1:name "Data space"@en ;
    schema1:url :dataSpace .

:termDenotation a schema1:DefinedTerm ;
    schema1:description "What a sign points to in the world — the “provides” role assigned to signs. The map–territory distinction is a claim about denotation: the sign is not the thing."@en ;
    schema1:inDefinedTermSet :glossaryTermSet ;
    schema1:isPartOf :article,
        :glossarySection,
        :glossaryTermSet ;
    schema1:name "Denotation"@en ;
    schema1:url dbr:Denotation .

:termDistributionalHypothesis a schema1:DefinedTerm ;
    schema1:description "The linguistic principle that words occurring in similar contexts have similar meanings — the theoretical ancestor of the contextual associations a language model learns."@en ;
    schema1:inDefinedTermSet :glossaryTermSet ;
    schema1:isPartOf :article,
        :glossarySection,
        :glossaryTermSet ;
    schema1:name "Distributional hypothesis"@en ;
    schema1:url dbr:Distributional_semantics .

:termExplicitSemantics a schema1:DefinedTerm ;
    schema1:description "Meaning identified by URIs, declared as RDF relationships, and shared via ontologies — machine-computable and open to logic."@en ;
    schema1:inDefinedTermSet :glossaryTermSet ;
    schema1:isPartOf :article,
        :glossarySection,
        :glossaryTermSet ;
    schema1:name "Explicit semantics"@en ;
    schema1:url :explicitSemantics .

:termImplicitSemantics a schema1:DefinedTerm ;
    schema1:description "Meaning encoded numerically and recoverable only by running the model — fluent and approximate, closed to deterministic inference."@en ;
    schema1:inDefinedTermSet :glossaryTermSet ;
    schema1:isPartOf :article,
        :glossarySection,
        :glossaryTermSet ;
    schema1:name "Implicit semantics"@en ;
    schema1:url :implicitSemantics .

:termLangulator a schema1:DefinedTerm ;
    schema1:description "The author's coinage for a large language model: a Language Emulation Machine. It emulates the systematic use of signs, syntax, semantics, and pragmatics without possessing language as a speaker does."@en ;
    schema1:inDefinedTermSet :glossaryTermSet ;
    schema1:isPartOf :article,
        :glossarySection,
        :glossaryTermSet ;
    schema1:name "Langulator"@en ;
    schema1:url :langulator .

:termPragmatics a schema1:DefinedTerm ;
    schema1:description "How situation fixes meaning: speaker, audience, time, place, purpose. The “bank” example — financial institution vs. river edge — is decided by pragmatics, which the model approximates as runtime context."@en ;
    schema1:inDefinedTermSet :glossaryTermSet ;
    schema1:isPartOf :article,
        :glossarySection,
        :glossaryTermSet ;
    schema1:name "Pragmatics"@en ;
    schema1:url dbr:Pragmatics .

:termSemantics a schema1:DefinedTerm ;
    schema1:description "What linguistic expressions mean — reference, sense, truth conditions. Implicit in the model (encoded associations); explicit in a Semantic Web (identified, declared relationships)."@en ;
    schema1:inDefinedTermSet :glossaryTermSet ;
    schema1:isPartOf :article,
        :glossarySection,
        :glossaryTermSet ;
    schema1:name "Semantics"@en ;
    schema1:url dbr:Semantics .

:termSign a schema1:DefinedTerm ;
    schema1:description "The perceptible carrier of meaning — a word, mark, or sound. In this mapping, signs become tokens and numerical representations inside the language model."@en ;
    schema1:inDefinedTermSet :glossaryTermSet ;
    schema1:isPartOf :article,
        :glossarySection,
        :glossaryTermSet ;
    schema1:name "Sign"@en ;
    schema1:url dbr:Sign_\(semiotics\) .

:termSyntax a schema1:DefinedTerm ;
    schema1:description "The combinatorial structure of a language: the rules by which signs combine into well-formed expressions. The model absorbs syntax as learned structural regularities, never as a declared grammar."@en ;
    schema1:inDefinedTermSet :glossaryTermSet ;
    schema1:isPartOf :article,
        :glossarySection,
        :glossaryTermSet ;
    schema1:name "Syntax"@en ;
    schema1:url dbr:Syntax .

:termToken a schema1:DefinedTerm ;
    schema1:description "The model's unit of text: sub-word pieces mapped to numerical vectors. Tokens are what signs become inside the Langulator — the raw material of all computation."@en ;
    schema1:inDefinedTermSet :glossaryTermSet ;
    schema1:isPartOf :article,
        :glossarySection,
        :glossaryTermSet ;
    schema1:name "Token"@en ;
    schema1:url dbr:Large_language_model .

:termTransformer a schema1:DefinedTerm ;
    schema1:description "The neural architecture (Vaswani et al., 2017) whose attention mechanism lets a model weigh context when computing representations — the machinery behind learned contextual association."@en ;
    schema1:inDefinedTermSet :glossaryTermSet ;
    schema1:isPartOf :article,
        :glossarySection,
        :glossaryTermSet ;
    schema1:name "Transformer"@en ;
    schema1:url dbr:Transformer_\(deep_learning\) .

:aboutSection a schema1:CreativeWork ;
    schema1:hasPart :kgGeneratorSkill,
        :museSparkAgent,
        :rdfInfographicSkill ;
    schema1:isPartOf :article ;
    schema1:name "About this collection"@en .

:langulator a schema1:Thing ;
    schema1:description "The essay's coinage for a large language model: a Language Emulation Machine. It emulates the systematic use of signs, syntax, semantics, and pragmatics without possessing language as a speaker does."@en ;
    schema1:isPartOf :article,
        :overviewSection ;
    schema1:name "Langulator"@en .

:sparqlWorkbenchSection a schema1:CreativeWork ;
    schema1:hasPart :recipeComponents,
        :recipeEntityTypes,
        :recipeFaqGlossary,
        :recipeHowto ;
    schema1:isPartOf :article ;
    schema1:name "SPARQL workbench"@en .

<https://kingsley.idehen.net/DAV/home/kidehen/Public/YouID/link-in-bio-credentials-5/index.html#netid> a schema1:Person ;
    schema1:description "The principal on whose behalf the collection was curated."@en ;
    schema1:name "Kingsley Idehen"@en .

:llmsLanguageOntology a owl:Ontology ;
    rdfs:label "LLMs and Language Ontology"@en ;
    schema1:description "Models the essay's argument: LanguageComponent instances mapping signs, syntax, semantics, and pragmatics onto LLM realizations; the Langulator concept; DivisionLayer instances for the five-layer division of labor; and the supporting FAQ, glossary, HowTo, image, and provenance entities."@en ;
    schema1:identifier <https://community.openlinksw.com/t/llms-and-language/6504> ;
    schema1:name "LLMs and Language Ontology"@en ;
    rdfs:comment "TBox for the map-territory story: language components, the Langulator, and the division of labor between implicit and explicit semantics."@en ;
    owl:versionInfo "1.0.0"@en .

:howtoSection a schema1:CreativeWork,
        schema1:HowTo ;
    schema1:description "Nine moves for working the essay's division of labor: begin and end with the human, let the Langulator handle the fuzzy parts, ground entities in RDF, Linked Data, and ontologies, compute the precise parts with logic, and keep the coupling between the two layers loose."@en ;
    schema1:isPartOf :article ;
    schema1:name "The division-of-labor pipeline"@en ;
    schema1:step :step1,
        :step2,
        :step3,
        :step4,
        :step5,
        :step6,
        :step7,
        :step8,
        :step9 .

:glossarySection a schema1:CreativeWork ;
    schema1:hasPart :glossaryTermSet,
        :termDataSpace,
        :termDenotation,
        :termDistributionalHypothesis,
        :termExplicitSemantics,
        :termImplicitSemantics,
        :termLangulator,
        :termPragmatics,
        :termSemantics,
        :termSign,
        :termSyntax,
        :termToken,
        :termTransformer ;
    schema1:isPartOf :article ;
    schema1:name "Glossary: the essay's vocabulary"@en .

:overviewSection a schema1:CreativeWork ;
    schema1:hasPart :compPragmatics,
        :compSemantics,
        :compSigns,
        :compSyntax,
        :divDataSpaces,
        :divHarness,
        :divLLM,
        :divLogic,
        :divSemanticWeb,
        :imageMapInfographic,
        :langulator,
        :relAttention,
        :relBert,
        :relDistributional,
        :relMeaning,
        :relMikolov,
        :relPeirce,
        :relRdfSemantics,
        :relShannon ;
    schema1:isPartOf :article ;
    schema1:name "Overview: the map and the territory"@en .

:glossaryTermSet a schema1:DefinedTermSet ;
    schema1:description "Twelve defined terms — standards-body and DBpedia IRIs where a canonical one exists, document-local fragments otherwise."@en ;
    schema1:hasDefinedTerm :termDataSpace,
        :termDenotation,
        :termDistributionalHypothesis,
        :termExplicitSemantics,
        :termImplicitSemantics,
        :termLangulator,
        :termPragmatics,
        :termSemantics,
        :termSign,
        :termSyntax,
        :termToken,
        :termTransformer ;
    schema1:isPartOf :article,
        :glossarySection ;
    schema1:name "Glossary: the essay's vocabulary"@en .

:faqSection a schema1:CreativeWork,
        schema1:FAQPage ;
    schema1:description "Twelve reader questions about the Langulator argument, answered from the source article."@en ;
    schema1:isPartOf :article ;
    schema1:mainEntity :faq1,
        :faq10,
        :faq11,
        :faq12,
        :faq2,
        :faq3,
        :faq4,
        :faq5,
        :faq6,
        :faq7,
        :faq8,
        :faq9 ;
    schema1:name "Frequently asked questions"@en .

:article a schema1:Article ;
    schema1:about dbr:SPARQL,
        :dataSpace,
        :explicitSemantics,
        :implicitSemantics,
        :langulator,
        <https://www.w3.org/2001/sw/#this> ;
    schema1:abstract "Language is a systematic use of signs, syntax, semantics, and pragmatics to encode and decode information — the map, not the territory. A large language model is therefore better understood as a Language Emulation Machine — a Langulator: signs become tokens and numerical representations, syntax becomes learned structural regularities, semantics becomes learned contextual associations, and pragmatics becomes runtime context. But an LLM's semantics are implicit and numerically encoded, while a Semantic Web makes them explicit — identified, declared, and machine-computable. This collection maps the four language components onto the model, contrasts implicit with explicit semantics, and lays out the division of labor: language for the fuzzy parts, a Semantic Web for the precise parts, loosely coupled between them."@en ;
    schema1:accountablePerson <https://kingsley.idehen.net/DAV/home/kidehen/Public/YouID/link-in-bio-credentials-5/index.html#netid> ;
    schema1:author <https://kingsley.idehen.net/DAV/home/kidehen/Public/YouID/link-in-bio-credentials-5/index.html#netid> ;
    schema1:hasPart :aboutSection,
        :compPragmatics,
        :compSemantics,
        :compSigns,
        :compSyntax,
        :dimCoupling,
        :dimFailureMode,
        :dimIdentity,
        :dimInference,
        :dimRepresentation,
        :dimSemantics,
        :divDataSpaces,
        :divHarness,
        :divLLM,
        :divLogic,
        :divSemanticWeb,
        :faq1,
        :faq10,
        :faq11,
        :faq12,
        :faq2,
        :faq3,
        :faq4,
        :faq5,
        :faq6,
        :faq7,
        :faq8,
        :faq9,
        :faqSection,
        :glossarySection,
        :glossaryTermSet,
        :howtoSection,
        :imageMapInfographic,
        :kgExplorerSection,
        :kgGeneratorSkill,
        :langulator,
        :museSparkAgent,
        :overviewSection,
        :rdfInfographicSkill,
        :relAttention,
        :relBert,
        :relDistributional,
        :relMeaning,
        :relMikolov,
        :relPeirce,
        :relRdfSemantics,
        :relShannon,
        :sparqlWorkbenchSection,
        :step1,
        :step2,
        :step3,
        :step4,
        :step5,
        :step6,
        :step7,
        :step8,
        :step9,
        :termDataSpace,
        :termDenotation,
        :termDistributionalHypothesis,
        :termExplicitSemantics,
        :termImplicitSemantics,
        :termLangulator,
        :termPragmatics,
        :termSemantics,
        :termSign,
        :termSyntax,
        :termToken,
        :termTransformer ;
    schema1:headline "LLMs and Language"@en ;
    schema1:isBasedOn <https://community.openlinksw.com/t/llms-and-language/6504> ;
    schema1:name "LLMs and Language"@en ;
    schema1:publisher dbr:OpenLink_Software ;
    prov:wasGeneratedBy <https://github.com/OpenLinkSoftware/ai-agent-skills/tree/main/kg-generator#this>,
        <https://github.com/OpenLinkSoftware/ai-agent-skills/tree/main/rdf-infographic-skill#this> .