Lexical Markup Framework: ISO24613

1
Lexical Markup FrameworkISO-24613

• Provo meeting
• Gil Francopoulo

2
schedule

• Brief history sum up where you are
• Future

3
Brief history and sum up

• We started in 2003
• The document is currently in DIS status. The NBs
  have until December to express their comments.
• We have to produce the FDIS for the end of
  February 2008
• LMF will be published in September 2008. With the
  help of AFNOR, we will produce a version in
  French.

4
Objectives

• LMF is a specification for interchange and
  representation of lexicons
• For MRD and NLP lexicons
• For all types of NLP applications
• For all languages

5
Structure of the document

• A document of 86 pages
• A good section on definitions (we spent a lot of
  time on this part)
• A core section based on UML
• 18 small annexes with many examples in many
  languages

6
Sum up of the model

• LMF specifies the structure of a lexicon
• All attribute adornment is made from data
  categories taken from the DCR
• LMF is defined by a UML specification for the
  classes and relations between the classes
• Many sub-parts are optional only the core
  package is mandatory

7
Core package
8
Various packages
9
Morphology representation in extension of the
morphology of the entries
10
Small example
11
Same data serialized in XML

• ltLexicalResource dtdVersion"14"gt    ltGlobalInfor
  mation        ltfeat att"languageCoding"
  val"ISO 639-3"/gt    lt/GlobalInformationgt
•     ltLexicongt
•     ltfeat att"language" val"eng"/gt
•     ltLexicalEntrygt            ltfeat
  att"partOfSpeech" val"commonNoun"/gt            
  ltLemmagt                ltfeat att"writtenForm"
  val"clergyman"/gt            lt/Lemmagt           
   ltWordFormgt
•                  ltfeat att"writtenForm"
  val"clergyman"/gt                 ltfeat
  att"grammaticalNumber"singular"/gt            lt/
  WordFormgt            ltWordFormgt
•                 ltfeat att"writtenForm"
  val"clergymen"/gt                ltfeat
  att"grammaticalNumber"plural"/gt            lt/Wo
  rdFormgt
•     lt/LexicalEntrygt    lt/Lexicongtlt/LexicalResour
  cegt

12
The future

• Looking for a successfull ISO standard
• STEP-1 define the specificationgt well advanced
• STEP-2 communicate
• STEP-3 Wed like LMF to be used

13
Some comments as clues for the future

• In a presentation in Tubingen this Spring, a
  person in the conference room asked me the model
  looks fine, well defined and powerful but, as a
  lexicographer its no use for me because I dont
  have any tool.

14
Provide a tool

• As a first version a simple tool, that (may be)
  will not implement the whole model. A stand alone
  version.
• Something free, open source

15
Provide other external formats

• Due to the fact that an ISO document is basically
  a text document of limited length, we could not
  produce some formats. And also because of time
  constraints.
• An RDF specification
• An ODD specification

16
Provide guidelines studies

• An LMF user guide with examples that we did not
  had space to insert in the LMF document
• A full technical study on how to map all famous
  lexicons in our field

17
Publish the data categories usable in an LMF
lexicons

• Taken from the three DCR profiles-
  morpho-syntax- syntax- semantics

18
Any other ideas ???

• We already have a list dedicated to LMF, but we
  could have a web site
• Keep on publishing