170829

Samest meeting 29.08.2017

Participants: Heiki, Heli, Jaak, Jack, Sjur, Trond

Agenda

The presentation

Highlights

Language technology actually needs linguists, not only statisticians.

We have made things that actually work, and that are on the net.

We have made learning programs for minority languages (Heli, Sulev, Jaska).

The work done on the keyboard infrastructure for Võro ś, ń, l´, t´, k´, h´

Debian has included hfst-based spelling ==> our work will give both Estonian and Võro speller for Debian

Narrative for the presentation

Perhaps more about the meaning of life, science, the universe and all that:

This paper presents infrastructure and language technology work for languages with a rich and productive morphology. The infrastructure (called Giella) may be characterised as a de- velopment environment for language processing. It contains language-independent modules for compiling and building both finite-state grammatical models and a wide range of language technology applications, as well as language-specific modules, one for each language in question. The infrastructure, grammatical models and language applications are all available as open source.

Intro with philosophy and all that

Quoting the application

  • Estonians: "have analysis, need infrastructure for applications"
  • Norwegians: "have infrastructure for applications, open for more languages to add"
  • Outcome: "Languages added, applications online, + new applications made for this project is reused for other languages as well"

Content and results

...

Future perspectives

Future perspectives in two parts:

- application-wise (what can we do that people woud like (to the extent that they contribute)

- content-wise (what is so fun and scientifically promising that we will do it as research)

Work:

  • FST development -- is going on, and Estonian gives new perspectives on FST manipulation
  • ICALL -- Tromsø now has a new programmer, Võro is going on + good prospects (educational partners?)
  • Speller -- student project to improve? -- new platforms: is continuing via the debian development, New languages, new techniques (weighting, context, mobile..., grammar checking)
  • MT -- is continuing -- langtech with an industry to back it up... (OmegaT with MT gets users?)

Applications for grants (Where, What)

Going to the conference

21-22 september Heiki; Sjur or Trond

Next meeting

Wednesday 13 September at 13:00 Norwegian time

Agenda:

  • presentation
  • LREC