Overview
SIST ISO 24613-2:2021 - part of the Lexical Markup Framework (LMF) - defines the Machine-Readable Dictionary (MRD) model, an implementation‑independent metamodel for representing electronic dictionary data. Building on the LMF core model (ISO 24613‑1), this part specifies UML classes, associations, data‑category selection and cross‑reference (CrossREF) principles to support a wide range of lexicon designs - from human‑oriented bilingual dictionaries to rigorously constrained lexica for machine processing.
Key topics and technical requirements
- Metamodel and class model: MRD is represented as UML classes and associations. Core classes reused from the LMF core include LexicalResource, Lexicon, LexicalEntry, Lemma, Form, Sense and Definition; new MRD classes include WordForm, Stem, WordPart, RelatedForm, Translation, Example, FormRepresentation, TextRepresentation, SubjectField, Bibliography.
- Class selection and multiplicity: Designers choose a subset of classes and set association multiplicities to match design goals (e.g., monolingual vs. bilingual MRDs). Optional classes have minimum cardinality zero.
- Generalization (typing): The model supports subclassing (e.g., Form → Lemma, WordForm, Stem, WordPart) to allocate specific data categories and constraints to subclasses.
- CrossREF model: Cross‑references link Forms, Senses and Translations, enabling rich interconnections across lexical entries (a capability emphasized in LMF Part 1).
- Data category selection: Annex A provides examples of data categories; developers may reuse or define domain‑specific data categories for features such as morphology, part of speech, usage notes.
- Object realization and serialization: The standard is implementation‑independent but discusses realization choices; commonly used serializations include XML and JSON, with tradeoffs in element/attribute modeling.
- Supporting morphology and MWEs: The MRD model supports extensional morphological descriptions and flexible modeling of multiword expressions (MWE) using CrossREFs and typed classes.
Practical applications and users
Who benefits:
- NLP engineers and computational linguists building lexicons for parsing, tagging, or machine translation.
- Machine translation and localization teams creating bilingual or multilingual MRDs.
- Lexicographers and dictionary publishers designing structured, interoperable electronic dictionaries.
- Ontology and terminology engineers linking lexical forms to semantic resources.
Practical uses:
- Designing interoperable lexicons for NLP pipelines.
- Creating bilingual dictionaries with explicit translation mappings.
- Modeling full morphological paradigms for language resources.
- Serializing lexicons in XML/JSON for dissemination and tool integration.
Related standards
- ISO 24613‑1 (LMF core model) - normative foundation for ISO 24613‑2.
- Other parts of the ISO 24613 series provide complementary LMF extensions and guidance for specific lexical resource types.
Keywords: SIST ISO 24613-2:2021, Lexical Markup Framework, LMF, Machine‑Readable Dictionary, MRD model, lexicon, CrossREF, UML, XML, JSON, morphology, bilingual dictionary, NLP, machine translation.
