UIT The arctic university of Norway > Giellatekno
 

Meeting_2010-09-20

Meeting setup

  • Date: 20.9.2010
  • Time: 09.30 Norw. time
  • Place: Internet
  • Tools: SubEthaEdit, iChat

Agenda

Cf. one of the following, depending on context:

  • the upper bar of the SEE window (provided you use the JSPWiki syntax mode)
  • the TOC in Forrest-rendered output, like HTML and PDF

Opening, agenda review, participants

  • Opened at 10: 03.
  • Present: Børre, Ciprian, Maja, Sjur, Thomas, Tomi, Trond
  • Absent: none
  • Agenda accepted as is

Updated task status since last meeting

Børre

  • get the 2006-2010 Sámi Parliament protocolls into our repository
    • ongoing
  • find and contact the correct person in SD, to get the manuscript for all Sámi teaching material now being given out to the schools
    • not done
  • fix the upload script after the corpus infra remake
    • not done
  • forrest makeover:
    • make the site build again
      • ongoing
    • upgrade all forrest installations, so we are all using the same baseline
      • not done
    • make the readme files for the proofing tools functional again
      • not tested
    • implement language switch for static divvun site
      • not done
  • contact the Sámi translators at the government and ministeries to get more parallell texts
    • not done
  • turn on corpus summary crontab
    • not done, Ciprian is reimplementing and improving corpus tools
  • improve XSL script to transform leaflet Forrest XDocs to an OOo Draw document
    • not done
  • get translations of thank-you letter
    • not done
  • make the new SL Server services functional:
    • group calendars
      • not done
  • fix bugs!

Ciprian

  • compile a db with parallel texts sme-nob for Autsch (deadline 20.09.2010)
    • ongoing work: too much unreliable meta-data in the xml files
  • refine the dict sorting script to indent like XXE
    • todo
  • update the XXE configuration to include the oahpa additions
    • done
  • 'Nachwehen' of the Sma-lex Workshop:
    • xml-ize all data from the Excel, Word, etc. files received from the different participants
    • enrich the data for the duodje terminology for users' need
      • todo
  • Kildin-Oahpa:
    • fix the localization bug (the sjd text doesn't show up)
      • todo
  • fitswe: check verbs from the csv-files and if needed put them into the xml format
    • todo
  • Skolt-Oahpa:
    • fine tuning of semantic tags
    • finish localization
      • todo
  • PhD-Projektbeskrivelse (internal deadline end of September 2010):
    • search, read, write (execute): high priority!
      • ongoing
  • update corpora (both monolingual and parallel) for Oslo (Glossa)
    • check the correctness of parallelity between sme and nob files
      • done: the script might need some refinment
    • sentence-align sme with nob: fix tca2 problem or search for a working solution (a different aligner)
      • found a sentence aligner for Hungarian-English: test needed
      • a small problem might be that it doesn't work on xml
    • analyse/disambiguate: todo -- waiting for the last version of FSTs
  • infrastructure
    • test cwb
    • continue with restructuring and cleaning the script catalogue as suggested in the newsgroup
      • todo
    • transform sme-lexC files into XML format
      • todo
    • make a schema/dtd description of the lexC-file (experiment with Komi and Romanian)
      • todo
  • GT web:
    • add a tree visualizer for the dependency trees
      • ongoing
    • add input help for special characters on the tool sites
      • todo
    • automatise the web statistics
      • todo
    • filter (English, German, etc.) input using language detection tools
      • todo
    • put a note on the sites that these are NOT MT tools
      • todo
    • input help for generating wordforms (dropdown menus).
      • todo
  • Sandbox Oahpa:
    • integrate reCAPTCHA into Django (as Lene suggested)
    • debug the installed sb_oahpa
    • update and correct the Oapha docu site
      • todo for all subpoints
  • Running Oahpa:
    • implement a testbench for Vasta (as Lene needs)
    • integrate reCAPTCHA after the SB-test
    • add an Oahpa clock and date excercise (cf. Numra)
      • todo for all subpoints
  • dictionaries, generally:
    • synchronize the source language entries from a specific dictionary with the entries in the morphology component (now especially for sma): that means nothing then put the entries from dict that are NOT analyzed into the lexC files
    • the StarDict on Windows: try the HTML-plugin (that means that users can use StarDict 3.2)
    • try to reduce the dict-size on mac: experiment with xPointer, etc.
  • Fkv: Nob - Nob: Fkv:
    • incorporate novel glossary into the dict
  • SmeNob:
    • incorporate the passives into the last version of the sme: nob
    • start a new compilation of SmeNob and improve it based on the experience gained with SmaNob
  • SmaNobSwe:
    • extend the smanobswe dictionary: waiting for data (incorporate the data from Maja-Lisa)
    • todo for all subpoints
  • MT
    • test A-ITE on Windows
      • todo, awaiting Windows machines.

Maja

  • more work on sma adjectives
    • working on it when I´m working with the loanwords - contemporary as the loanwords
  • go through the lexicon and add for all loan words a form corresponding to the suggestion from Ove Lorentz (other forms should be SUB-marked)
    • ongoing
  • continue gathering sma corpus texts
    • SMS från Soppero (130 sider)
      • Contacted Ajlin Jonetta JOnassen - positive
      • Sent mail and waiting an answer from Laestadius
      • Contacted Podium -positive
  • finish compound tags for adjectives
    • ongoing
  • Rette goldstandard doc - så mange som mulig - high prio! - min. 1000 (helst 5000) ord løpende tekst
    • not prio right now
  • test conversion to xml to ensure there are no markup errors in goldstandards
    • not started yet
  • continue gathering sma corpus texts
    • prio
    • get sma articles in Š-bláđđi
      • sent mail to Iđut
    • the Gun Utsi book is almost there - one contract missing
      • give up - send an e-mail to Ristin Turi Aleksandersen and Julie Eira about this -
  • Corpus gathering:
    • contact Inga Margrethe Bjørn Eira
      • Why should I contact here? You have to go back to earlier meeting memos and check
    • follow up the yearbook-articles
      • they were aall positive, need to start scanning - send them contracts
    • restart the letter mailing thing using WebSak
      • Don't know what this is; you should send out letters to the authors and publishers that the contract is now registered, and under which archive number in the Sámi Parliament. This has not been properly done for all contracts in the past, but Børre started to do this earlier. Thus, he is the one to give you an introduction.OK!
  • adjectives
    • working on it when I´m working with the loanwords - contemporary as the loanwords
  • fix bugs!
    • not done

Sjur

  • update the XXE configuration to include the oahpa additions
    • done, version 1.2 available but not announced (need to update the Wiki to include a description of how to update an XXE config first)
  • write sma beta readme file
    • not yet done
  • forrest makeover:
    • discussion with Børre
    • make the site build again
    • upgrade all forrest installations, so we are all using the same baseline
    • make the readme files for the proofing tools functional again
  • convert M4 twolc processing to transducer manipulation
    • discussed with Tomi
  • add voikko support to our proofing test bench
    • not yet
  • add all our Sámi analysers to voikko and test them as spellers
    • not yet, but doing hfst-based speller experiments using hfst-ospell
  • run tests using Hunspell, Voikko, Polderland for our Sámi lexicons
  • difftest for fst and PL speller
  • Northern areas
    • plan a meeting/seminar in Tromsø
    • make a plan for the first two years, update overall plan
  • @TTS: continue public tender process
  • make Leif Åge send out CD's to distribution points
  • continue Nordplus Sprog project
    • see hfst-ospell work
  • make XSL script to transform leaflet Forrest XDocs to an OOo Drawer document
  • follow-up on some Polderland-related bugs: 621, 630, 652
  • fix bugs!
  • other things:
    • preparations for OCR speller discussions at NB

Thomas

  • go through sma loanwords and change according to Trondheim findings
    • worked and still working
  • write sma readme file
    • not done
  • check extra hyphens in the lexc continuations with Trond
    • briefly checked
  • Digitalize south saami books
    • nothing this week
  • fix bugs!
    • some work done

Tomi

  • new sma beta on Friday, September 17; compilation to start Thursday 17: 00
    • done, but tests showed that something is not right
  • run tests using Hunspell, Voikko, Polderland for our Sámi lexicons
    • not done
  • put together the TTS preprocessing transducers and scripts
    • not done
  • try to compile voikko
    • nothing this week
  • add all our Sámi analysers and test them as spellers
    • not done
  • convert M4 twolc processing to transducer manipulation
    • done, except hyphenation - working with makefile
  • write sma beta readme file
    • not done
  • document how compounding is controlled in the PLX conversion
  • fix double hyphen bugs
  • fix PL smj hyphenator bug
  • fix PL conversion bugs
  • fix bugs!

Trond

  • check extra hyphens in the lexc continuations with Thomas
    • Nothing new here
  • Course plan for Barents meeting/seminar, october. Financing must be in place.
    • Nothing new on Barents.
    • check the time needed to tet visas with the UiT admin
    • Nothing new on Barents.
    • plan a meeting/seminar in Tromsø
    • Nothing new on Barents.
    • make a plan for the first two years, update overall plan
  • MT/Terminology
    • Done some on MT, nothing on terminology
  • sma-oahpa: text analysis and setup of basic vocabulary
    • Done.
  • Follow up the issue of Windows test machines with IT dept
    • Two machines are clear (XP, Vista), but not received, awaiting the third (W7)
    • Waiting for info from IT dept as to when
  • New corpus interface for Oslo (with Ciprian)
    • We are not there yet.
  • fix bugs!.

Oahpa!

Trond has done a lot of translations to Swedish. Needs to be checked, and is not complete.

TODO

  • From the start: sjd_oahpa Leksa in deu and eng as well (Ciprian)
  • improve sjdoahpa (deadline 13. September 2010) (Ciprian):
  • Get
    • add/correct Leksa entries
    • fix the localization bug (the sjd text doesn't show up)
    • delayed
  • sma-oahpa:
    • Mark curriculum words from candidate list, in the smanob dictionary ( Ciprian)
    • Work on the list (Røros)
      • Røros gang has started working
  • clock and date for Numra (Ciprian)
  • add Captcha for the feedback e-mail address (Ciprian)

Corpus gathering

National Library: there is a meeting September 23, to which Trond and Sjur are going. Trond will request a couple of Sámi OCR-ed texts (together with the nob texts we already have), and use them as test cases for spell-checking and (automatically) correcting the OCR-ed texts.

What we need now is free North Sami - Norwegian parallel text, because of Autshumato.

TODO:

  • get the 2006-2010 Sámi Parliament protocolls into our repository (Børre)
  • contact the Sámi translators at the government and ministeries to get more parallell texts (Børre)
  • continue gathering sma corpus texts (Maja)
    • get sma articles in Š-bláđđi
    • the Gun Utsi book is almost there - one contract missing (Maja)
  • send a copy of the signed contracts back to the authors, translators and publishers, accompanied by the thank-you letter
  • find and contact the correct person in SD, to get the manuscript for all Sámi teaching material now being given out to the schools (Børre)
  • get the sma yearbooks from Saemien sïjhte ( Maja)
  • contact Inga Margrethe Bjørn Eira (Maja)
  • give Maja a WebSak intro (Børre)
  • restart the letter mailing thing using WebSak (Maja)

Promoting Divvun

TODO:

  • make leaflet to inform about the project (Børre)
    • add InDesign text (Sjur)
    • make XSL script to transform Forrest XDocs to an OOo Drawer document ( Børre, Sjur)
  • distribute CD version through the library bus, the language centres and common sami centres in all of Sápmi. Gaaltije in Östersund for example. ( Leif Åge, Sjur)
    • make him send out CD's accordingly (Sjur)
  • update online download log statistics page (Børre)
  • make an info campaign about our language technology products and services (financed by the Sámi Parliament) - send out people that travel around and demonstrate and inform about our tools etc.

Future plans, directions and ideas

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

Northern areas project

First major obstacle: make working keyboards and fonts. In principle, there are two approaches to keyboards:

  1. Make the Russian-based keyboards work
  2. Design new Kildin-optimized keyboards

The first goal has priority. We should have text analysis in order to consider whether the second goal is feasible. Trond to follow up test machines.

  • Choice of fonts
  • Rendering of fonts in MS Word and other important programs
  • Keyboard layout
    • Empirical phase: What keyboards are around
    • Design phase: Make the optimal keyboards -- for all OSes

Write trustworthy and detailed documentation (in Russian)

What we know:

  • you can type on some computers (Michael R: it works), not on others. Q: what are the differences between the two types of computers?
  • some fonts can't display the chars, others can (but usually ugly)

TODO:

  • Course plan for meeting/seminar, october. Financing must be in place.
    • check the time needed to tet visas with the UiT admin (Trond)
    • call the ministry (Trond)
  • Attend a beginners' course in Russian (priority: the alphabet!) near you..
  • make a plan for the first two years, update overall plan (Trond, Sjur)
  • Follow up Windows test machines

Infrastructure

Out of the box experiences:

  • fao could not be compiled out of the box (the Oslo gang tried) because of a missing tmp dir
    • is easily resolved by a mkdir -p command in the make files
  • our list of software to install is incomplete or lacking installation instructions
  • external tools require manual installation
  • there are bugs in the gtsetup.sh script

Updated corpus online

See Ciprian´s document about the corpus content in $GTPRIV/plan/corpus/oslo_corpus_update_todo.txt.

Issues:

  • filenames
  • organisation of subdirs?
  • content (or lack thereof) of original and converted files
  • svn repo reorganisation
facta$convert2xml.pl --nolog --corpdir=/usr/local/share/corp L1allOrt.correct.txt

Error message:

sh: /home/sjur/gtmain/gt/script/text_cat: No such file or directory
L1allOrt.correct.txt: ERROR errors in /home/sjur/gtmain/gt/script/text_cat -q \
   -x -d /home/sjur/gtmain/gt/script/LM "/usr/local/share/corp/tmp/L1allOrt.correct.txt.tmp0": 

text_cat isn't part of our repository, we need to add it - we are using a modified version of the original file. It should be added with a short README and some license info, pointing to the original.

TODO:

  • make the conversion scripts work with the new svn repository ( Børre, Ciprian)
    • add check for almost empty content
      • already implemented in convert2xml.pl. Still empty or almost empty files are converted. This is because of unexpected structure of content, so that the conversion goes wrong, and leaves no real content in the resulting xml file.
    • add processing of svg files
      • not done
  • reorganise subdirs as needed (Ciprian, Trond)
    • not yet done
  • identify parallel nob files - should be automatic, but needs to be checked ( Ciprian)
  • sentence-align sme with nob (Ciprian)
    • the aligner wasn't working, Børre has tried to fix it
      • not yet fixed - bug not found
  • analyse/disambiguate (Ciprian)
  • preferably also dep -> check with Oslo if it can be used (Ciprian)
  • install in Oslo (Ciprian, Trond)
  • add better handling of unknown strings in our analysers (???)

Corpus infra remake

TODO:

  • meeting to go through open issues and divide workload Tuesday 10.00 ( Børre, Ciprian, Sjur, Trond)
  • fix the upload script after the corpus infra remake (Børre)
  • turn on corpus summary crontab (Børre)
  • needs to check the conversion from old rcs corpus repository to new svn-based repository (Ciprian)

Corpus interface

This depends on the infrastructure cleanup.

TODO:

  • make a simple web search form for the UiT corpus repository (Ciprian)
  • check out the new version of CWB with Unicode support (Ciprian)
    • fall

Makefile + tag simplification

First part done. All linguists need to test the new transducer, to see that we get what we want. Most important: that the descriptive / normative border is as it used to be. Tomi has been testing norm/desc fst's with xfst minus operator. Now we also need to compare actual output from the two fst's with before and after versions.

TODO:

  • convert M4 twolc processing to transducer manipulation (Tomi, Sjur)
    • first do norm/desc merging
      • done - needs (more) testing
    • then do hyphenation
  • testing (only sme):
    • make new fst's (Tomi)
    • make old fst's (Thomas)
    • analyse text with old and new, descr. and norm - compare results ( Thomas, Lene, Trond)

HFST compilation

Our HFST transducers do still contain all the extra tags resulting from the tag conversion (cf above). The Apertium versions do not, however. The main problem here is the forking of smi transducer builing and maintenance - one branch in Giellatekno, and another in Apertium. This is not good for future maintenance, and should be corrected ASAP. The tag-removing make steps should be incorporated in the giellatekno svn.

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:

  • forrest makeover: ( Børre, Sjur)
    • make the site build again
    • upgrade all forrest installations, so we are all using the same baseline
    • make the readme files for the proofing tools functional again
    • implement language switch for static divvun site
  • make the new SL Server services functional: ( Børre)
    • group calendars
  • finish the restructuring and cleaning of the script/ directory Ciprian)
  • infrastructure remake: ( Børre, Ciprian, Sjur, Tomi, Trond)
    • more modularised make / build infra (prepare for smn, sms, sjd, others)
      • look at omorfi for ideas of how to modularise
    • merge gt, kt and st into one
    • modularised preprocess and spellrelax
    • alternatives to make:
  • make a test-all target that runs all tests we have (Ciprian, Sjur, Trond)
    • delayed until we have restructured the make/build process
  • define and document testing routines (Ciprian, Sjur, Trond)
    • delayed until we have restructured the make/build process

Linguistics

North Sámi

There are a lot of hyphens found in noun paradigms. They should not be there. Trond and Thomas will look into them.

TODO:

  • check extra hyphens in the sme lexc continuations (Thomas, Trond)

Lule Sámi

TODO:

  • check extra hyphens in the smj lexc continuations (Thomas, Trond)

South Sámi

We shall first convert the loanwords in the existing lexicon, before considering adding more loanwords.

TODO:

  • go through the lexicon and add for all loan words a form corresponding to the suggestion from Ove Lorentz (other forms should be SUB-marked) ( Maja, Thomas)
  • adjectives (Maja with Thomas, Trond, Sjur)
    • two competing naming conventions of continuation lexicons
      • One naming goes ATTRSUFF-PREDSUFF-STEMTYPE
      • One follows the sme convention of naming key adjectives
      • There are duplicate lexica
    • The comparative issue open here and there
      • The ATTRSUFF-PREDSUFF-STEMTYPE lexica now go to EVENCOMP and ODDCOMP only.
  • finish compound tags for adjectives (Maja)

Name lexicon/risten.no infrastructure

Most of the items that used to be listed here are now moved to the new risten2 project, scheduled for next year.

TODO:

  • refine the requirements for risten.no 2.0 (Ciprian, Sjur, TermTeam, SD)

Dictionaries

Ciprian has made a new online/offline dictionary interface based on Apertium tools. See http://victorio.uit.no/webdict/gtd_lookup.html.

WeDict dictionary client for iOS (uses stardict dictionary files): http://app.weiphone.com/wedict/

iPod/iPhone has (almost) no problem with Unicode text input (but users HAVE to know how to use unicode as input) There are two letters missing, ŧ, ŋ, the other ones are there. You may use Serbo-Croat, or QUERTY, when writing North Sami on iPod.

There is one possible issue: http://code.google.com/p/wedictpro/issues/detail?id=2

StarDict is useless for Cyrillic languages, mainly because of the scanning function (ie point and look-up). We need to find an alternative to StarDict. Also, the Kildin Sámi users have different needs than we have had in mind. And they don't have access to a Kildin Sámi keyboard in Windows.

  • we need a Russian Windows machine, bought in Russia, using the software they use
  • we also of course need a western/Norwegian Windows machine for testing

Released:

  • SME: NOB update release

Other things dictionary-related:

  • risten.no data as part of our dictionaries
  • dictionaries as part of risten.no
  • dictionaries and risten.no as part of Autshumato ITE

TODO:

  • kom: fin-eng
    • moved the original kom-lex.xml to the inc-dir and froze it
    • split it by pos into the working_file dir, the ONLY place to work with the dictionary entries
    • now, the lexC files are generated via XSLT sheets, no perl scripts
    • adjusted the Makefile
    • prepared the pipeline for compiling the mac dict
    • todo: make a pipeline for StarDict also (as far as I know, Jaska has) a Linux machine
  • set up risten.no on eXist/XServe (as a beta version site) (Sjur)
  • set up required infra for smenob on risten.no/XServe (Sjur)
  • Continue the dictionary infrastructure discussion (Ciprian, Sjur, Trond)
  • end user documentation (how to download and install) (Ciprian, Trond)
  • Contact Davvi Girji about cooperation on electronic dictionaries ( Sjur)
  • developing the mobile phone version of smenob:
  • Komi
    • take out the doublets to a separate file (Ciprian)
    • merge the doublets (Jaska, Trond)
    • Completing the automaton to some state (Trond, Jaska, Paula)
  • make the sort XSL script available for all languages to keep the source files sorted

Proofing tools

Spelling feedback from Malta:

  • use FST to model suggestions (Krister et al)
  • Lene is evaluating the output of the Polderland speller
    • first-letter errors rare, but the PLX speller still changes it quite often
    • Oahpa and Divvun approaching each other
  • using Wikipedia as a source for spelling errors and test material (a French group has made a setup of tools to automatise most of this)

South Sámi

TODO:

  • test conversion to xml to ensure there are no markup errors ( Maja, Sjur, Thomas)
  • difftest for fst and PL speller (Sjur)
  • write readme file (Sjur, Thomas, Tomi)
  • new beta version on Friday, September 17; check-in deadline on Thur (Tomi)

HFST- and Voikko-based proofing tools

TODO:

  • run tests using Hunspell, Voikko, Polderland for our Sámi lexicons ( Sjur, Tomi)
  • check out the voikko code, see this page or this page ( Tomi)
  • try to compile Voikko (development is done on Linux, no MacOS X testing so far) (Tomi)
  • add voikko support to our proofing test bench (Sjur)
  • add all our Sámi analysers and test them as spellers (Tomi, Sjur)

Speller bugs

List of bugs returned from Polderland:

  • 621
  • 630
  • 652
  • 656
  • 676

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

  • 399 - missing numerals (plural forms) - still OPEN
  • 425 - X not recognised; single letters were left out - still OPEN
  • 435 - roman numbers - inflection of single letter numbers rejected, as well as some complex numbers (but is ok in smj) - still OPEN
    • we should pregenerate all numbers once and for all, and store them in a separate lexicon file
  • 461 - REGRESSION: missing suggestion (sáhkki)
  • 508 - REGRESSION: accepts smj entries (most likely abbreviation missing)
  • 520 - REGRESSION: r9 and š9 not defined (abbr. missing)
  • 595 - prefix+name without hyphen (ovdaLot instead of ovda-Lot) - still OPEN
  • 603 - suomabealdi accepted - still OPEN
  • 606 - compound-tags LEXICON VUOHTA - still OPEN
  • 613 - short gen. as second compound part - still OPEN
  • 619 - numerals and pronouns to NAMÁK and SASJ fails - vihttasoarttat remaining - still OPEN
  • 629 - a taking part in compounding without hyphen - still OPEN
    • only open case has word A-finálaid compounded
  • 647 - numerals+NOUN - still OPEN, open case has uppercase letters
  • 648 - unmotivated suggestions with numeral+noun - still OPEN
  • 661 - REGRESSION: abbr. not recognized
  • 709 - sámedikkeválga accepted - OPEN
  • 728 - vowel shortening GenCmp+Left-tagged - still OPEN
  • 779 - caseforms of pronoun okatahat - still OPEN
  • 785 - does not recognize alphabet-abbr+noun - OPEN
  • 802 - NEW: multiword propernouns
  • 803 - NEW: FINJU- words accepted single-handed
  • 804 - NEW: guovttilogát, njealjilogát
  • 805 - NEW: Nouns+acronyms

smj

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

  • 435 - roman number - single letter numbers now recognised
    • we should pre-generate all numbers once and for all, and store them in a separate lexicon file
    • please note that inflection of single letter numerals is fine in smj, as opposed to sme
  • 482 - polardutkamin not recognized - FIXED
  • 496 - REGRESSION: unrecognised clitics
  • 556 - non-existent word accepted - FIXED
  • 594 - lågenanguoktáj not recognized - still OPEN
  • 595 - REGRESSION: prefix+name as split comp without hyphen
  • 596 - C-giellan is not accepted - still OPEN
  • 600 - Gen+hyph compound - FIXED
  • 627 - prefix + hyhpen does not get accepted - FIXED
  • 647 - numerals+NOUN - still OPEN, open case has uppercase letters
  • 648 - unmotivated suggestions with numeral+noun - still OPEN
  • 650 - REGRESSION: noun prefix+name compound without hyphen
  • 652 - UPPERCASE-typos only get acronym-suggestions - still OPEN
  • 692 - numeral-variants - all but one fixed (gáktsalågenantjuotakta), but still OPEN
  • 744 - REGRESSION: numerals + clitic
  • 803 - NEW: VINJU- words accepted single-handed
  • 805 - NEW: Nouns+acronyms

TODO:

  • document how compounding is controlled in the PLX conversion (Tomi)

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

  • 670 - Hard hyphen replaced with soft hyphen: 10-biejvvásattja (the word is not rec.; Bug #711) - still OPEN

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:

  • fix PL hyphenator errors (Tomi)
    • almost done - one smj bug left

Installer changes

TODO:

  • test InDesign installer (Sjur)

User documentation

TODO:

  • InDesign documentation (Sjur)
    • Norwegian translation received from Davvi Girji

1.2 release

Content:

  • several smj bug fixes
  • lexicalisations
  • InDesign Mac & Win
  • new OOo beta
  • improved installers, at least for Mac, preferably also for Windows

2.0 release

Content:

  • official sma proofing tools release

Nordplus Språk proofing test bench project

Arno Teigseth is hired for the summer. He is sitting in Equador, 7 hours after Norway. He is working with a quechua speller (using hunspell) and dictionary.

A goal is to use texts from parallel domains, in order to be able to compare nob and sme. We start out with 100000 words, 5000 for each genre:

  1. Discussion fora on the net
  2. Blogs

Discussion fora can be problematic, as they contain a lot of very oral language with deliberate oral or dialectal forms used in the text. As such, the texts won't be typical for the kind of texts people would like to spell-check, and thus the genre is problematic for the purpose at hand.

Possible other genres:

  1. Minutes from local associations
  2. Pupil texts - look for the home pages of the schools
  3. exam texts by 10th grade pupils, where the pupils have published the texts themselves
  4. skrivebua - Nordland fylkeskommune: all three Sámi languages

Text to speech

There will be regular meetings in this project from now on, every second week.

TODO:

  • put together the preprocessing transducers and scripts (Tomi)
  • refine syntax / dependency rules (Biret Ánne)
  • continue public tender process (Sjur)

Machine Translation

Background:

Kevin and Ryan have now closed the work on smenob and finsme. They will continue, but mainly do other things. As for smesmj, the possible student input is rescheduled for October.

We hope to have a student working one month on smesmj this summer. The world now knows (vaguely) that we work on MT, so we should do, as well. We also need to work together with the external ones.

All languages

TODO:

  • Improve the sme-dis for MT purposes
  • Go through the bilingual transfer lexica for more general translations
  • Add alternative translations and selection rules
  • Arrange an Apertium seminar (later)
  • transfer hfst build infra from Apertium to Giellatekno

finsme

Technical problems with compunds in Finnish: each part in the compound is separated by a space character (+Ux20), which is creating problems in the transfer.

TODO

  • Tag unification
  • Adp split in fin (Trond)

smenob

Needed: Lexicon completion. Grammatical transfer rules

smesmj

New worker underway.

smesma

Infrastructure in place, to demo in Trondheim to encourage more work and projects.

CAT

Ciprian has made a translation memory based on our parallel corpus. The test looks promising, he has identified certain issues. We need to be clear about the license of the original parallel texts, and we need to know the original language (ie translation direction).

TODO:

  • make intelligent nobsme risten.no - requires nob morphology
    • nob.fst a la Trond
    • norskordbank.fst
    • nob.fst based on an xml source synchronised with Ordbanken source texts
    • should be coordinated with the Apertium analyser
  • write a TM script to turn a parallell corpus into TM entries (Ciprian)
    • deadline: next Monday, 20.9.
  • testing A-ITE on Norwegian Windows
    • done by our friends in RSA - bug identified, workaround sent by e-mail

SMA seminar in August/September followup

TODO:

  • update the XXE configuration to include the oahpa additions ( Ciprian, Sjur)
    • done but not yet announced, needs user documentation first
  • write user documentation (Sjur)

Other

Start to make yearly reports

Directory in svn in place, now we only need to fill it with content...

Thursday inhouse seminar

Topic for this week: hfst + spelling + ocr (Sjur, perhaps Lene)

Next time suggestion list:

  1. introduction to xslt - Ciprian to start out
    1. relevant xslt issues:
    2. basic principles of xslt …
    3. sorting in xslt … (have a look at the dictionary sort xslt script)
    4. converting from one xml format to another wilt xslt (sugg: convert from DivvunGT dictionary dtd to MacDict xml)

Future seminars:

  1. XQuery
  2. More XML (needs concretisation)
  3. UML
  4. other suggestions?

Fall planning

Topics:

  • speller test project -> Sjur, Børre, Thomas, X
  • speech synthesis -> Sjur, Trond, BA as a starter
  • risten.no -> Ciprian, Sjur, Tomi
  • Barents project follow-up meeting -> Trond, Sjur
  • sme and smj proofing tools, next version -> Thomas, Tomi, Sjur
  • HFST-based proofing tools -> Sjur
  • MT terminology project -> Trond, Linda, Fran, Kevin
  • Lexicography seminar follow-up (evt. jan-feb)
  • Participate in Sámi center sje-project -> gt

Dates:

  • September 22-23: Mo i Rana/Nasjonalbiblioteket/OCR
  • October 6.-10. RiLiVS conference in Reykjavík (topic: corpora for Skandinavan languages) (Ciprian)
  • October 11.-15. sma-oahpa kickoff (sma-oahpa group)
  • October 27-28: Linköping seminar on multilinguality and accessibility ( Sjur)
  • November 1st: Noregsuniversitetet sme-oahpa2 application deadline
  • Desember: Divvun release
  • March/April 2011: full-scale lexc course in Budapest (Sjur)

Next meeting, closing

The next meeting is 27.9.2010, 09: 30 Norwegian time.

The meeting was closed at 10: 46.

Appendix - task lists for the next week

Boerre

iCal

  • get the 2006-2010 Sámi Parliament protocolls into our repository
  • find and contact the correct person in SD, to get the manuscript for all Sámi teaching material now being given out to the schools
  • fix the upload script after the corpus infra remake
  • forrest makeover:
    • make the site build again
    • upgrade all forrest installations, so we are all using the same baseline
    • make the readme files for the proofing tools functional again
    • implement language switch for static divvun site
  • contact the Sámi translators at the government and ministeries to get more parallell texts
  • turn on corpus summary crontab
  • improve XSL script to transform leaflet Forrest XDocs to an OOo Draw document
  • get translations of thank-you letter
  • make the new SL Server services functional:
    • group calendars
  • fix bugs!

Ciprian

iCal

  • compile a db with parallel texts sme-nob for Autsch (deadline delayed because of the unreliable meta-data)
  • refine the dict sorting script to indent like XXE
  • 'Nachwehen' of the Sma-lex Workshop:
    • xml-ize all data from the Excel, Word, etc. files received from the different participants
    • enrich the data for the duodje terminology for users' need
    • if needed, compile a StarDict version for the client wedict/wedictpro for iPhone/iPod
    • Trond and/or Sjur should test the client on their hardware
  • Kildin-Oahpa:
    • fix the localization bug (the sjd text doesn't show up)
  • fitswe: check verbs from the csv-files and if needed put them into the xml format
  • Skolt-Oahpa:
    • fine tuning of semantic tags
    • finish localization
  • PhD-Projektbeskrivelse (internal deadline end of September 2010):
    • search, read, write (execute): high priority!
  • update corpora (both monolingual and parallel) for Oslo (Glossa)
    • check the correctness of parallelity between sme and nob files
    • sentence-align sme with nob: fix tca2 problem or search for a working solution (a different aligner): high priority
    • analyse/disambiguate: todo -- waiting for the last version of FSTs
  • infrastructure
    • test cwb
    • continue with restructuring and cleaning the script catalogue as suggested in the newsgroup
    • transform sme-lexC files into XML format
    • make a schema/dtd description of the lexC-file (experiment with Komi and Romanian)
  • GT web:
    • add a tree visualizer for the dependency trees
    • add input help for special characters on the tool sites
    • automatise the web statistics
    • filter (English, German, etc.) input using language detection tools
    • put a note on the sites that these are NOT MT tools
    • input help for generating wordforms (dropdown menus).
  • Sandbox Oahpa:
    • integrate reCAPTCHA into Django (as Lene suggested)
    • debug the installed sb_oahpa
    • update and correct the Oapha docu site
  • Running Oahpa:
    • implement a testbench for Vasta (as Lene needs)
    • integrate reCAPTCHA after the SB-test
    • add an Oahpa clock and date excercise (cf. Numra)
  • dictionaries, generally:
    • synchronize the source language entries from a specific dictionary with the entries in the morphology component (now especially for sma): that means nothing then put the entries from dict that are NOT analyzed into the lexC files
    • the StarDict on Windows: try the HTML-plugin (that means that users can use StarDict 3.2)
    • try to reduce the dict-size on mac: experiment with xPointer, etc.
  • Fkv: Nob - Nob: Fkv:
    • incorporate novel glossary into the dict
  • SmeNob:
    • incorporate the passives into the last version of the sme: nob
    • start a new compilation of SmeNob and improve it based on the experience gained with SmaNob
  • SmaNobSwe:
    • extend the smanobswe dictionary: waiting for data (incorporate the data from Maja-Lisa)
  • MT
    • test A-ITE on Windows

Maja

iCal

  • more work on sma adjectives
  • go through the lexicon and add for all loan words a form corresponding to the suggestion from Ove Lorentz (other forms should be SUB-marked)
  • continue gathering sma corpus texts
  • finish compound tags for adjectives
  • Rette goldstandard doc - så mange som mulig - high prio! - min. 1000 (helst 5000) ord løpende tekst
  • test conversion to xml to ensure there are no markup errors in goldstandards
  • continue gathering sma corpus texts
    • get sma articles in Š-bláđđi
    • the Gun Utsi book is almost there - one contract missing
  • Corpus gathering:
    • contact Inga Margrethe Bjørn Eira
    • follow up the yearbook-articles
    • restart the letter mailing thing using WebSak
  • adjectives
  • fix bugs!

Sjur

iCal

  • document updating XXE configurations; announce the upgrade
  • write sma beta readme file
  • forrest makeover:
    • make the site build again
    • upgrade all forrest installations, so we are all using the same baseline
    • make the readme files for the proofing tools functional again
  • convert M4 twolc processing to transducer manipulation
  • add voikko support to our proofing test bench
  • add all our Sámi analysers to voikko and test them as spellers
  • run tests using Hunspell, Voikko, Polderland for our Sámi lexicons
  • difftest for fst and PL speller
  • Northern areas
    • plan a meeting/seminar in Tromsø
    • make a plan for the first two years, update overall plan
  • @TTS: continue public tender process
  • make Leif Åge send out CD's to distribution points
  • continue Nordplus Sprog project
  • make XSL script to transform leaflet Forrest XDocs to an OOo Drawer document
  • follow-up on some Polderland-related bugs: 621, 630, 652
  • fix bugs!

Thomas

  • go through sma loanwords and change according to Trondheim findings
  • analyse text with old and new, descr. and norm - compare results
  • make old fst's for testing M4 removal in makefile
  • write sma readme file
  • check extra hyphens in the lexc continuations with Trond
  • Digitalize south saami books
  • fix bugs!

Tomi

  • new sma beta on Friday, September 17; compilation to start Thursday 17: 00
  • run tests using Hunspell, Voikko, Polderland for our Sámi lexicons
  • put together the TTS preprocessing transducers and scripts
  • try to compile voikko
  • add all our Sámi analysers and test them as spellers
  • convert M4 twolc processing to transducer manipulation
  • write sma beta readme file
  • document how compounding is controlled in the PLX conversion
  • fix double hyphen bugs
  • fix PL smj hyphenator bug
  • fix PL conversion bugs
  • fix bugs!

Trond

iCal

  • analyse text with old and new, descr. and norm - compare results
  • check extra hyphens in the lexc continuations with Thomas
  • Course plan for Barents meeting/seminar, october. Financing must be in place.
    • check the time needed to tet visas with the UiT admin
    • plan a meeting/seminar in Tromsø
    • make a plan for the first two years, update overall plan
  • Follow up the issue of Windows test machines with IT dept
  • fix bugs!.