ISO 24613-2

SLOVENSKI STANDARD oSIST ISO/DIS 24613-2:2019 01-oktober-2019 Upravljanje jezikovnih virov - Ogrodje za označevanje leksikonov (LMF) - 2. del: Model za strojno berljiv slovar (MRD)

Language resource management -- Lexical markup framework (LMF) -- Part 2: Machine

Gestion de ressources linguistiques -- Cadre de balisage lexical -- Partie 2: Titre manque

2003-01.Slovenski inštitut za standardizacijo. Razmnoževanje celote ali delov tega standarda ni dovoljeno.

All rights reserved. Unless otherwise specified, or required in the context of its implementation, no part of this publication may

be reproduced or utilized otherwise in any form or by any means, electronic or mechanical, including photocopying, or posting

on the internet or an intranet, without prior written permission. Permission can be requested from either ISO at the address

Foreword ........................................................................................................................................................................................................................................iv

Introduction ..................................................................................................................................................................................................................................v

1 Scope ................................................................................................................................................................................................................................. 1

2 Normative references ...................................................................................................................................................................................... 1

3 Terms and definitions ..................................................................................................................................................................................... 1

4 Key standards used by LMF ....................................................................................................................................................................... 1

5 The machine readable dictionary (MRD) model ................................................................................................................. 1

5.1 General ........................................................................................................................................................................................................... 1

5.2 MRD class model ................................................................................................................................................................................... 2

5.2.1 Set of classes ........................................................................................................................................................................ 2

5.2.2 Class selection and cardinality............................................................................................................................. 2

5.2.3 Generalization .................................................................................................................................................................... 3

5.2.4 Object realization ............................................................................................................................................................ 3

5.3 Data category selection and class population .............................................................................................................. 3

5.4 CrossREF allocation ............................................................................................................................................................................ 3

5.5 Form subclasses ..................................................................................................................................................................................... 4

5.5.1 WordForm class ................................................................................................................................................................ 4

5.5.2 Lemma class ......................................................................................................................................................................... 4

5.5.3 Stem class ............................................................................................................................................................................... 4

5.5.4 WordPart class ................................................................................................................................................................... 4

5.5.5 RelatedForm class ........................................................................................................................................................... 4

5.6 FormRepresentation class ............................................................................................................................................................. 4

5.7 TextRepresentation class ............................................................................................................................................................... 5

5.8 Translation class .................................................................................................................................................................................... 5

5.9 Example class ........................................................................................................................................................................................... 5

5.10 SubjectField class .................................................................................................................................................................................. 5

5.11 Bibliography class ................................................................................................................................................................................ 5

5.12 Multiword Expression (MWE) Analysis ............................................................................................................................ 5

Annex A (informative) Data category examples ....................................................................................................................................... 7

Annex B (informative) Machine readable dictionary examples ............................................................................................... 9

Bibliography .............................................................................................................................................................................................................................22

ISO (the International Organization for Standardization) is a worldwide federation of national standards

bodies (ISO member bodies). The work of preparing International Standards is normally carried out

through ISO technical committees. Each member body interested in a subject for which a technical

committee has been established has the right to be represented on that committee. International

organizations, governmental and non-governmental, in liaison with ISO, also take part in the work.

ISO collaborates closely with the International Electrotechnical Commission (IEC) on all matters of

The procedures used to develop this document and those intended for its further maintenance are

described in the ISO/IEC Directives, Part 1. In particular, the different approval criteria needed for the

different types of ISO documents should be noted. This document was drafted in accordance with the

editorial rules of the ISO/IEC Directives, Part 2 (see www .iso .org/directives).

Attention is drawn to the possibility that some of the elements of this document may be the subject of

patent rights. ISO shall not be held responsible for identifying any or all such patent rights. Details of

any patent rights identified during the development of the document will be in the Introduction and/or

Any trade name used in this document is information given for the convenience of users and does not

For an explanation of the voluntary nature of standards, the meaning of ISO specific terms and

expressions related to conformity assessment, as well as information about ISO's adherence to the

World Trade Organization (WTO) principles in the Technical Barriers to Trade (TBT), see www .iso

This document was prepared by Technical Committee ISO/TC 37, Language and terminology,

This second edition cancels and replaces together with parts 1, 3, 4, 5, and 6 the edition ISO 24613:2008,

This edition merges two normative annexes from the previous edition, Annex A, Morphology extension,

and Annex C, Machine readable dictionary extension, providing a more cohesive description of the

key structures (classes and associations) found in that edition. The cross-reference (CrossREF) model

introduced in Part 1, Core model, of this edition, provides a new capability for correlating lexical

features across different form and sense classes. In addition, the CrossREF model has replaced the

ListOfComponents and Component classes, enabling a more extensible and flexible capability for

managing multiword expressions. The meta-model of generalization by typing introduced in Part 1

provides a more rigorous and unambiguous framework for applying LMF modelling mechanisms in

ways that enable greater editorial freedom and support the comparison of different LMF conformant

designs. This edition has kept most of the informative examples found in the previous edition (deleting

only a few redundant examples) and has added new examples to illustrate new modelling features.

There have been some class name changes (e.g. OrthographicRepresentation for Representation and

Translation for Equivalent), but no changes in the underlying concepts of the previously existing classes.

Any feedback or questions on this document should be directed to the user’s national standards body. A

The ISO 24613 (LMF) series is based upon the definition of an implementation-independent meta-model

combining a core model and additional models that onomasiological (form-oriented) and semasiological

LMF also provides guidelines for various implementation use cases, and where appropriate describes

This part extends ISO 24613-1, the LMF Core model, through the use of the processes and mechanisms

described in ISO 24613-1. The objective is to enable flexible design methods to support the development

of machine readable dictionaries for different purposes while enabling cross comparisons of different

designs and a basis for developing assessments of standards conformance. The scope of supported

design goals ranges from simple to complex human-oriented MRDs, both monolingual and bilingual,

lexicons that support conceptual-lexical systems through links with ontological resources, rigorously

constrained lexicons for supporting machine processes, and lexicons that provide an extensional

This International Standard describes the Machine readable dictionary (MRD) model, a metamodel for

representing data stored in a variety of electronic dictionary subtypes, ranging from direct support for

human translators to support for machine processing. Since this part of the ISO 24613 series is based on

ISO 24613-1, the LMF Core model, it is designed to interchange data with other parts of the ISO 24613

The following documents are referred to in the text in such a way that some or all of their content

constitutes requirements of this document. For dated references, only the edition cited applies. For

undated references, the latest edition of the referenced document (including any amendments) applies.

ISO 24613-1, Language resource management — Lexical Markup Framework (LMF) — Part 1: Core model

For the purposes of this document, the terms and definitions given in ISO 24613-1 apply.

ISO and IEC maintain terminological databases for use in standardization at the following addresses:

The key standards applicable to this part are described in ISO 24613-1, the LMF Core model.

The MRD model is represented by UML classes, associations among the classes (the structure), sets

of data categories (attribute-value pairs), and links (cross-references). The following sections describe

The classes defined in ISO 24613-1, the LMF Core model, that are used in the MRD extension include

LexicalResource, GlobalInformation, Lexicon, LexiconInformation, GrammaticalInformation,

LexicalEntry, Lemma, Form, Sense, Definition, OrthographicRepresentation, and principles

for applying the CrossREF class. These classes, together with the associations and constraints

described in ISO 24613-1, are applicable to the design of MRD. New classes introduced in this part

include the WordForm, Stem, WordPart, RelatedForm, Translation, Example, FormRepresentation,

The sets of classes shown in the model in Figure 1 can support a wide range of design objectives. A

specific design objective can require all or only some of the classes shown in the above model and

can require as well the creation of new subclasses. The recommended first step in the creation of a

model for a specific design objective (e.g. a bilingual dictionary) should be the selection and possible

exclusion of classes contained in the class model and the application of desired cardinalities to the

class associations as required by the model and the design goals (the optional classes in the model

have a minimum cardinality of zero). The developer can create new subclasses, as needed, using the

mechanisms described in ISO 24613-1, the LMF Core model. The selected classes and their associations

will provide the structure and nodes (classes) appropriate for the intended lexical design. The classes

— Certain classes of MRD, such as monolingual and bilingual dictionaries, generally require a Sense class

— Certain classes of MRD, such as concept hierarchies, may not require a Form class instantiation.

— Certain classes of MRD, such as orthographic dictionaries and extensional morphologies may not require a

— Certain classes of MRD, such as the extensional morphologies, can provide constraints on the attributes

NOTE The purpose of the MRD morphology extension is to provide the mechanisms to support the

development of lexicons that have an extensional description of the morphology of lexical entries in which all

Figure 1 illustrates the use of generalization (typing) through the Form class (superclass) and its

subclasses, the Lemma, WordForm, Stem, and WordPart, and the OrthographicRepresentation

(superclass) and its subclasses, the FormRepresentation and TextRepresentation. The typing

mechanism describes how to allocate specific sets of data categories, associations, cardinalities, and

cross-references to subclasses (e.g. Lemma) in order to redefine the superclass. ISO 24613-1 provides a

NOTE The subclasses shown in Figure 1 are available for use in LMF compliant designs, but are not

exhaustive, since LMF allows the creation of additional subclasses. The lexicon designer specifies what sets of

LMF provides examples of object models (see Annex B), but does not provide an in-depth description

of the overall methodologies for developing the object models, since those processes are heavily

dependent on the choice of model serialization (e.g. XML, JSON). Different serializations can require

different design approaches and impose limitations on how the object can be modelled.

EXAMPLE XML provides a number of structural models for implementing XML schemas. Within the

framework of these models, a lexicon designer could implement UML classes as XML elements or a combination of

an XML element and attributes. For example, a designer could instantiate the Lemma class as a element

or a

and data category allocations to implement object designs that are strongly dependent on the structures and

Data category selection can include all or a subset of data categories used by a given domain. Examples

of data categories and their allocations are listed in Annex A. Where needed, the lexicon developer can

Figure 1 shows links (cross-references) between the Form and Sense and the Form and Translation

classes. The principles for modelling cross-references are described in ISO 24613-1, the LMF Core model.

The CrossREF class is specifically allowed for the LexicalEntry class, the Lemma class, WordForm class,

WordPart class, Sense class, and the Sense class children. The lexicon designer should consider using

cross-references with the RelatedForm. The use of data categories to provide information about the

CrossREF features (e.g. internal reference, external reference, type of ID, lexical type, syntactic type, or

EXAMPLE A WordPart that contains the suffix component of a Lemma can be cross-referenced with the

LexicalEntry that contains that suffix as the Lemma, or a Sense can be cross-referenced with a broader Sense

contained in a different LexicalEntry, or an authentic Quote can be cross-referenced with a document that

WordForm is a Form subclass containing a word form, such as an inflected form, that a lexeme can take

when used in a sentence or a phrase. The WordForm class is in a zero-to-many aggregate association

with the LexicalEntry class (inheriting the Form cardinality). The WordForm class can manage simple

lexemes, compounds, multi-word expressions, and sub-lexemes such as affixes and roots.

Lemma is a Form subclass representing a lexeme or sub-lexeme used to designate the LexicalEntry

(part of the Form-Sense paradigm). The Lemma class is in a zero-to-one aggregate association with the

LexicalEntry that overrides the multiplicity inherited from the Form class (see ISO 24613-1 for a more

Stem is a Form subclass containing a stem or root. The Stem class can be typed as a specific type of

stem or root (e.g. type=”arabicRoot”). The Stem is in a zero-to-one aggregate association with the

WordPart is a Form subclass representing sub-lexeme parts other than the stem or root (e.g. affix,

prefix, suffix). The WordPart is in a zero-to-many aggregate association with the LexicalEntry.

RelatedForm is a Form subclass containing a word form or a morph that is typical of run-on entries

in print dictionaries. The RelatedForm has a different Sense than the Lemma and can be considered

a candidate for eventual inclusion in a different Lexical entry. The RelatedForm can be related to

the Lemma in a variety of ways (e.g. synonym, cross-reference, multi-word expression, idiom). The

RelatedForm is in a zero-to-many aggregate association with the LexicalEntry and can contain a

cross-reference to a different lexical object. The RelatedForm can be typed (generalization) using data

EXAMPLE A developer possibly wants to use the RelatedForm for a multi-word expression (e.g. United

States) that contains a component form of a Lemma (e.g. united). The design goal could be to preserve the format

of the original source material, or to provide immediate user support while developing an improved lexicon that

FormRepresentation is an OrthographicRepresentation subclass that contains the text literal and

metadata (e.g. pronunciation, hyphenation, xml: lang, script) for a Lemma, WordForm, or other subclass

of the Form class. FormRepresentation is in a zero-to-many aggregate association with a Form subclass.

NOTE Data categories, such as xml: lang, script, and notation, are associated with the

EXAMPLE Because searching for WordPart data (e.g. suffix components of a form) is generally not a high

user priority, a lexicon developer might want to create a PartRep subclass of the FormRepresentation class

in order to support application designs that use object (class) names as part of their query strategy. Creating

different search criteria for FormRepresentation objects and PartRep objects is one way to increase search and

TextRepresentation is an OrthographicRepresentation subclass that manages the text literals and

metadata (e.g. xml: lang, script) for classes associated with the Sense class and its child classes. The

TextRepresentation class enables improved modelling of the Form-Sense paradigm by distinguishing

the management of descriptive text literals in the Sense related classes from text literals that represent

a form (e.g. word form, sub-lexeme). The TextRepresentation class is in a zero-to-many aggregate

association with the Definition, Translation, or Example class. The TextRepresentation class can be

NOTE In practice, a TextRepresentation class associated with a Definition class will most likely be in a one-

EXAMPLE A Quote subclass could be created for a TextRepresentation class associated with an Example

In a bilingual MRD, the Translation class represents the translation equivalent of the word form

contained by the Lemma or WordForm class. The Translation class is in a zero-to-many aggregate

association with the Sense class, which allows the lexicon developer to omit the Translation class from

The Example class contains a text string that illustrates the usage of the Lemma, WordForm, or

Translation in authentic or constructed context. The Example class can be typed (subclass) in order to

further define the context. The Example class is in a zero-to-many aggregate association with the Sense

class. The Example class can also be in a zero-to-many aggregate association with the Translation class

when there is a need to differentiate among multiple Example objects contained in a Sense, not all of

SubjectField is a class managing subject domain for a Sense. The SubjectField class is in a zero-to-

many aggregate association with the Sense class. Based on the design goals, it is possible that a lexicon

developer wants to constrain the cardinality of the SubjectField. When there are multiple Senses, each

Bibliography is a class containing source information for a LexicalEntry or an object (class instatiation)

of the Lexical Entry. The Bibliography can be associated with more than one class in a LexicalEntry,

NOTE The Bibliography class can more typically be associated with a Form subclass, the Sense, or one of the

LMF enables an analysis of MWE using cross-references between an instantiation of a Lemma subclass

containing the MWE and two or more different entries, each of which contains a component form of the

MWE. Possible targets can include the Lemma objects or the Sense objects in the other entries. A design

that targets the Sense objects would be the optimum design choice for reducing semantic ambiguity.

The LexicalEntry objects can be in the same lexicon, a different lexicon, or even a different resource. The

PartOfSpeech and other grammatical features describing the component word forms can be different

NOTE In the most common use cases for MWE analysis, the MWE is contained in the Lemma, but it is possible