Language Technology at UiT

The Divvun and Giellatekno teams build language technology aimed at minority and indigenous languages

View GiellaLT on GitHub divvungiellatekno/giellalt.uit.no

Meeting setup

Agenda

Cf. one of the following, depending on context:

Opening, agenda review, participants

Opened at 10:00.

Present: Børre, Ciprian, Maja, Sjur, Thomas, Tomi, Trond

Absent: none

Agenda accepted as is.

Updated task status since last meeting

Børre

Ciprian

Maja

Sjur

Thomas

Tomi

Trond

Oahpa!

A lot of work for the Bodø version, but MorfaC needs more work.

Now all the Oahpa dev is done in $GTPRIV, it should be done in $GTHOME. The move might be delayed till after the Bodø seminar.

A swedish-speaking guy would like to do the LexC translations into Swedish.

Meeting memos can be found at [http://giellatekno.uit.no/ped/index.html#Meeting+memos]

TODO

Corpus gathering

Børre talked to Ávvir two years ago. They had changed their publishing system to an Adobe-based one which doesn’t give direct access to articles as individual files. Now the articles are stored in a database.

We should contact Ávvir to find out how to get text cooperation working again, both regarding their article database (corpus) and their user dictionaries for the spellers.

We also need to tighten the routines regarding the SD protocols, so that we get all new protocols regularly.

Which books are digitized?

TODO:

Promoting Divvun

TODO:

Future plans, directions and ideas

See a separate document in plan/strat/5year.jspwiki.

Northern areas project

TODO:

Infrastructure

Makefile + tag simplification

TODO:

  1. convert comment tags to real lexc tags when the discussion is over (Tomi)
    1. done
  2. adjust gt/Makefile to grep on the new tags such that the old make targets work (Tomi)
    1. done (works without changes)
  3. add the new tags to the tags-not-save.regex file (this will remove them from the regular analyzer) => the transducers and their output should be identical to the earlier versions (Tomi)
    1. done, but not tested
  4. write new build commands (Sjur, Tomi)
  5. when the new build infrastructure works as it should, delete the old ones (Sjur, Tomi)

General list

To accommodate future enhancements in different directions (in rough order of importance):

  1. test bench for all parts of our language technology efforts
    1. test bench enhanced, but not yet complete
  2. improve Forrest i18n support with static sites
  3. reorganise the documentation:
    1. differ between target groups
    2. get better grouping
    3. decide what to write in Forrest and what in wiki (cf. Apertium and [http://xixona.dlsi.ua.es/apertium/]) for a similar split)
    4. update/add missing parts
  4. migrate lexc lexicons to XML, splitting the task
    1. Name lexica (the Name project)
    2. Dictionaries (already in XML, task is to integrate them)
    3. At least migrate the lexc open POSes (Komi as a pilot case)
  5. change the look of the documentation web
  6. corpus content moved to Max Planck repositories? Norsk språkbank?
  7. update infrastructure to allow content-restricted spellers for special target groups

TODO:

Linguistics

Hyphenation of loan words: Thomas would like to have a meeting about this with Maja, Trond, Lene. He will send an e-mail about it.

North Sámi

(nothing new, see proofing bugs below)

Lule Sámi

(nothing new, see proofing bugs below)

South Sámi

Many words do not not follow Umlaut-rules:

Analyser output:

håjjedh+V+Ind+Prs+Pl3 (jamre, klage)
hejjieh
håjjoeh (- > hujjieh, according to Maja)
håjjah

jmf: these are according to Umlaut-rules:

Umlaut in derived verbs in adjectives does not not follow Umlaut-rules

Lexc solution:

  1. one contlex for umlautverbs (for each class)
  2. one contlex for nonumlautverbs
  3. one contlex for directing to both (or merging to both) (for verbs with variation)

The third contlex could also be used to monitor behaviour, with tags +Uml +Nouml.

Numerals:

Three types of problem areas:

We also seem to have problems with derivations:

jealadehtedh   jieliedehtedh	jieledh+V+Der1+Der/dehte+V+Inf
galmadehtedh   galmedehtedh	galmedidh+V+Der2+Der/ehte+V+Inf

About the sma school conference: we’ll ask them about their promise from last year, and decide what to do based on their response. Initial decision is to go there.

TODO:

Name lexicon/risten.no infrastructure

Tomi has played with couchdb as a replacement for eXist in risten.no and general dictionary-related work. Seems much lighter and easier to work with.

Instead of building our own webforms and back-end update scripts, use XForms with a premade connection to our xml db. Orbeon XForms is such a tool (open source).

From the meeting with the terminology and IT teams last week:

This means the following tasks:

TODO:

  1. send eXist log files to Ciprian (Sjur)
    1. done
  2. fix i18n bug in risten.no/G5 (so they will work without the proper locale request) (Sjur)
  3. fix bugs in lexc2xml; add comments to the log element (Saara)
  4. finish first version of the editing (Sjur)
  5. test editing of the xml files. If ok, then: (Sjur, Thomas, Trond)
  6. make terms-smX.xml <=== automatically from propernoun-sme-lex.xml (add nob as well) (the morphological section should be kept intact, in e.g. propernoun-sme-morph.txt) (Sjur, Saara)
  7. convert propernoun-($lang)-lex.txt to a derived file from common xml files (Sjur, Tomi, Saara)
  8. implement data synchronisation between risten.no and the cvs repo, and possibly other servers (ie the G5 as an alternative server to the public risten.no - it might be faster and better suited than the official one; also local installations could be treated the same way)
  9. start to use the xml file as source file
  10. clean terms-sme.xml such that all names have the correct tag for their use (e.g. @type=secondary) (Thomas, linguists)
  11. merge placenames which are errouneously in different entries: e.g. Helsinki, Helsingfors, Helsset (linguists)
  12. publish the name lexicon on risten.no (Sjur)
  13. add missing parallel names for placenames (linguists)
  14. add informative links between first names like Niillas and Nils (linguists)

Dictionaries

All Oahpa! now - no dictionaries:)

fkv:nob and nob:fkv is scheduled to be released at March 18th. The doublet-removal work is ongoing (appr. 200 left), thereafter 225 pos=x should be added, so there might be more doublets.

What will be released is a two-way dictionary without morphology. The nice-to-have feature would be a web interface in addition to the StarDict and MacDict ones, but really as an extra, eventual bonus only.

TODO:

Proofing tools

South Sámi

TODO:

HFST-based proofing tools

The work with Voikko+HFST is slowly moving forward.

Testing

Spelling Error Markup

TODO:

Speller testing

TODO:

Testing open-source Norwegian spellers

Sjur has invited the open-source group to test their spell-checker using our test bench. The response has been positive, we’ll see what happens.

We should go to their developer meetings, and present our work and how to work with language technology.

Speller bugs

List of bugs returned from Polderland:

Tag reordering for abbreviations have caused a lot of problems:

smj:
hr.
hr.	hr+ABBR+Acc
cand.philol.
cand.philol.	cand.philol+ABBR+N+Acc
Per
Per	Per+N+Prop+Mal+Sg+Attr

sme:
hr.
hr.	hr+N+ABBR+Acc
Per
Per	Per+N+Prop+Mal+Sg+Attr

Open issues based on test results:

sme

Version: Davvisámi, version 1.2, 2009-09-18

smj

Version: Julevsáme, version 1.2, 2009-09-20

TODO:

Hyphenator bugs

Open issues based on test results :

sme

Lexicon version: Davvisámi, version 1.2, 2009-09-18

No known issues!

smj

Lexicon version: Julevsáme, version 1.2, 2009-09-20

sma

Command to test the hyphenator:

preprocess dev/corp/pressemelding.txt | lookup bin/hyph-sma.fst | cut -f2 | \
lookup bin/hyphrules-sma.fst | grep -v '^$' | cut -f2 | uniq | see

TODO:

Installer changes

TODO:

User documentation

TODO:

1.2 release

Content:

Other

Thursday inhouse seminar

A short (less than 1h) seminar every Thursday at 10 AM. Possible topics:

Spring planning

Topics:

Dates:

Text to speech

We need to put the preprocessing pieces together, ie all the bits that will turn regular texts into decent input for the synthesizer.

Files:

  1. sme/src/sme-num.txt for numeral to text
  2. … for abbreviation to text (not done(?))
  3. sme/src/phon-sme.xfst for text to IPA/Sampa

Present output of sme-num.fst:

3
3	golbma

23.2.
23.2.	 guovvamánu guoktelogigoalmmát beaivi
23.2.	guoktelogigolbma čuokkis nubbi

23.02.2010
23.02.2010	 guovvamánu guoktelogigoalmmát beaivi guokteduhátlogi

10:32
10:32	logi kolon golbmalogiguokte

We would like to have POS tagging as well, and perhaps intonation grammar (focus phrase detection)

First POS tagging and text processing The last step: conversion to Sampa.

TODO:

CAT

Autshumato ITE won’t work on the Mac because of Apple software security policies. It still works fine on Windows and Linux, and could possibly be used at the Sámi Parliament.

Next meeting, closing

The next meeting is 22.2.2010, 9.30 Norwegian time.

Next week: Ciprian is in Berlin next week, and returning on the following Sunday. Sjur is taking winter holidays, but will try to work in the afternoons.

The meeting was closed at 11:58.

Appendix - task lists for the next week

Boerre

Ciprian

Maja

Sjur

Thomas

Tomi

Trond