User Tools

Site Tools


Working Group 4: Quantifying and promoting diversity


This WG is transversal to WGs 1-3 and will focus on how the Action serves inter- and intra-linguistic diversity. Its activities will overlap with the 3 other WGs in:

  1. Networking for diversity:
    • bringing together pre-existing groups dedicated to NLP-applicable universality,
    • integrating experts of (notably low-resourced) languages not yet covered by these groups,
    • integrating experts in linguistic typology;
  2. Quantifying diversity:
    • designing measures of inter- and intra-linguistic diversity in language resources and tools,
    • using these measures to quantify diversity in UD and PARSEME corpora;
  3. Promoting diversity:
    • procedures for better use of the existing resources, based on their estimated diversity,
    • selecting new data to be annotated, so as to favour intra-linguistic diversity,
    • designing evaluation scenarios which favour tools performing well on rare and diverse phenomena and low resourced languages,
    • integrating and training new experts dedicated to low-resourced and endangered languages,
    • validating the unified annotation guidelines (WG1) and lexicon formats (WG2) against newly included languages and defining new language-specific categories and extensions, if needed,
    • coordinating of the creation and enhancement of annotated corpora and lexica for low-resourced languages,
    • discovering and analysing rare linguistic phenomena, and describing them in resources and tools,
    • coordination of the development of NLP tools (WG3) for low-resourced and endangered languages.


Translations of this page:
  • en
wg4/wg4.txt · Last modified: 2023/03/22 11:57 by agata.savary