@amlhubs/nif — NIF 2.0 as a Typed Metamodel

Identity

Field	Value
Standard	NLP Interchange Format (NIF) 2.0
Specification	NIF 2.0 Core
Ontology	nif-core OWL
Namespace IRI	http://persistence.uni-leipzig.org/nlp2rdf/ontologies/nif-core#
Authority	W3C Best Practices for Multilingual Linked Open Data Community Group (BP-MLOD CG) — academic hosting at AKSW/InfAI, Universität Leipzig
Edition	NIF 2.0 (frozen). NIF 2.1 RC is in extended draft at nif.readthedocs.io; 2.0 is the stable freeze.
npm Package	@amlhubs/nif
npm Version	0.0.1
Peer Dependencies	@amlhubs/uml, @amlhubs/mof, @amlhubs/ontolex
License	UNLICENSED

Authority Caveat

NIF 2.0 was developed inside the W3C BP-MLOD Community Group but never advanced to a chartered W3C Working Group nor reached W3C Recommendation status. The live specification is hosted at persistence.uni-leipzig.org, an academic URL maintained by the AKSW research group at InfAI / Universität Leipzig. A NIF 2.1 Release Candidate has been "in revision" on nif.readthedocs.io for several years without resolution.

This package is published per an explicit user override expanding the source-round scope despite the prior /metamodel source decision (.work/2026-05-03-source/02-decisions.md §R1) flagging the authority concerns. Consumers SHOULD weigh that governance posture against the practical interoperability the substring-offset URI scheme provides. If the Leipzig hosting lapses, every URI minted under the NIF namespace will silently break — a custom URI template under a project-controlled domain is a viable fallback.

Abstract

The NLP Interchange Format is an RDF/OWL ontology for interchange of natural-language-processing annotations across heterogeneous tools. NIF 2.0's load-bearing contribution is a deterministic substring-offset URI scheme based on RFC 5147 (#char=begin,end), which lets two independent NLP pipelines refer to the same character range of the same source text without coordinating on entity identifiers in advance. Around that URI scheme NIF defines a small class hierarchy — Context (the full source text), String (any character range over a Context), and the linguistic specializations Word, Phrase, Sentence, Title, Paragraph — together with positional and structural properties (beginIndex, endIndex, anchorOf, referenceContext, superString, subString, nextWord, previousWord, nextSentence, previousSentence, …) and an Annotation class for typed third-party annotations attached to a String.

The @amlhubs/nif npm package repackages the NIF 2.0 ontology as extensible TypeScript interfaces and base classes, citing the persistence.uni-leipzig.org section anchors in JSDoc above every metaclass. Each NIF concept is grounded in the AML upstream stack: Context and String extend the OntoLex-Lemon LexicalEntry shape so that NIF strings interoperate with lemmatization and lexical-resource pipelines; the offset URI scheme is encoded as a typed value object so the #char=begin,end fragment is parseable round-trip. The package also surfaces NIF's integration point with W3C ITS 2.0 (itsrdf:taIdentRef, the per-span text-analytics identifier reference) so NIF annotations and ITS-driven localization metadata can travel together over the same RDF graph.

Scope — What the Package Surfaces

The package targets the NIF 2.0 Core ontology surface, organized by section anchor on persistence.uni-leipzig.org/nlp2rdf/ontologies/nif-core/nif-core.html:

NIF Class	NIF-Core Anchor	Role
Context	nif:Context	The full source text — the anchor for every String URI minted within it
String	nif:String	Any character range over a Context, identified by #char=begin,end
Word	nif:Word	Token-level String (tokenization output)
Phrase	nif:Phrase	Multi-word constituent String (parser output)
Sentence	nif:Sentence	Sentence-boundary String (sentence segmenter output)
Title	nif:Title	A String representing a document/section title
Paragraph	nif:Paragraph	A paragraph-level String
Annotation	nif:Annotation	Third-party typed annotation attached to a String
OffsetBasedString	nif:OffsetBasedString	URI-scheme realization of String using RFC 5147 #char= fragments
ContextHashBasedString	nif:ContextHashBasedString	URI-scheme realization of String using context-hash digests

Plus the property surface:

NIF Property	Carrier	Codomain	Purpose
nif:beginIndex	String	xsd:nonNegativeInteger	Inclusive start offset (UTF-16 code units per spec §URI Schemes)
nif:endIndex	String	xsd:nonNegativeInteger	Exclusive end offset
nif:anchorOf	String	xsd:string	The literal characters at [beginIndex, endIndex)
nif:referenceContext	String	Context	The Context this String is offset within
nif:isString	Context	xsd:string	The full text content of a Context
nif:superString	String	String	Containing String (e.g., a Sentence containing a Word)
nif:subString	String	String	Inverse of superString
nif:nextWord	Word	Word	Sequential next Word in reading order
nif:previousWord	Word	Word	Inverse of nextWord
nif:nextSentence	Sentence	Sentence	Sequential next Sentence
nif:previousSentence	Sentence	Sentence	Inverse of nextSentence
itsrdf:taIdentRef	String	IRI	Text-analytics identifier reference (W3C ITS 2.0 integration point)

Dependency Topology

@amlhubs/nif is a downstream metamodel — it consumes UML for its metaclass machinery, MOF for reflective access, and OntoLex-Lemon as the lexical-resource grounding for Context/String.

@amlhubs/uml         (UML 2.5.1 — root)
   ▲
   │
@amlhubs/mof         (MOF 2.5.1 — reflective ring on UML)
   ▲
   │
@amlhubs/ontolex     (W3C OntoLex-Lemon — lexical-entry vocabulary)
   ▲
   │  peerDependencies (uml + mof + ontolex)
   │
@amlhubs/nif         (THIS PACKAGE — NIF 2.0 substring-offset URI scheme + nif-core hierarchy)

Installation & Usage

npm install @amlhubs/nif @amlhubs/ontolex @amlhubs/mof @amlhubs/uml

import type {
  IContext,
  IString,
  IWord,
  IAnnotation,
} from '@amlhubs/nif';

const ctx: IContext = {
  elementId: 'doc-1',
  isString: 'The quick brown fox jumps over the lazy dog.',
  beginIndex: 0,
  endIndex: 44,
};

const word: IWord = {
  elementId: 'doc-1#char=4,9',
  beginIndex: 4,
  endIndex: 9,
  anchorOf: 'quick',
  referenceContextId: ctx.elementId,
};

Provenance & Formal References

• NIF 2.0 Core Specification — the persistence-URL freeze of NIF 2.0
• nif-core ontology (HTML view)
• nif-core ontology (OWL/RDF source)
• W3C Best Practices for Multilingual Linked Open Data CG (host community group)
• AKSW / InfAI — Universität Leipzig (technical maintainer)
• NIF 2.1 RC (extended draft, non-normative)
• RFC 5147 — URI Fragment Identifiers for the text/plain Media Type — the source of the #char=begin,end substring-offset scheme
• W3C ITS 2.0 — itsrdf:taIdentRef integration point

Version History

Version	Date	Change Summary
0.0.1	initial publish	NIF 2.0 Core surface — Context, String, Word, Phrase, Sentence, Title, Paragraph, Annotation, OffsetBasedString, ContextHashBasedString + offset/anchor/superString/subString/nextWord/previousWord/nextSentence/previousSentence/itsrdf:taIdentRef property surface

License

UNLICENSED — restricted npm access under @amlhubs scope at npm.pkg.github.com.