UIT The arctic university of Norway > Giellatekno
 

Meeting_2006-09-25

Meeting setup

  • Date: 25.09.2006
  • Time: 09.30 Norw. time
  • Place: Where we are
  • Tools: SubEthaEdit, iChat

Agenda

  1. Opening, agenda review
  2. Reviewing the task list from two weeks ago
  3. Documentation - divvun.no
  4. Corpus gathering
  5. Corpus infrastructure
  6. Infrastructure
  7. Linguistics
  8. name lexicon infrastructure
  9. Spellers
  10. Other issues
  11. Summary, task lists
  12. Closing

1. Opening, agenda review, participants

Opened at 09: 32.

Present: Børre, Sjur, Thomas, Tomi, Trond

Absent: Maaren, Saara

Agenda accepted as is.

2. Updated task status since last meeting

Børre

  • corpus collection:
    • send out contracts with accompanying letter
    • Gather public texts, preferrably also parallel ones
    • Send out letters to the rest of the Iđut authors
    • contact Ája (Kåfjord), talk to Lene
    • send contracts to Čálliid Lágádus
    • contact Richard Valkepää at NSI about older Min Áigi and Áššu files
    • discuss with Bård Eriksen about collecting smj texts (with Sjur)
      • Asked him to send us a book catalogue, so that we can contact authors.
  • corpus conversion:
    • convert nob and nno bible texts to be used as part of a parallel corpus
    • convert fin, swe to paratext or directly to our XML
    • review the paratext2xml converter
    • Move norwegian documents in Min Áigi from sme to nob
  • corpus access:
    • possibly deploy the user account form as an HTML form
    • Write both user and admin documentation (Børre, review: Sjur, Thomas)
      • User documentation probably in several languages. This covers how to apply for an account, on what grounds one can apply, and pointers to documentation telling how to use the corpus.
      • Admin documentation, telling how we set the permissions to the corpus files, and whatever other processes and tasks needed to set up a corpus account.
  • set up Bugzilla automatic reminders for open issues
  • create document & document entry for semantic double-tagging of names (for Trond)
  • finish Forrest i18n and Sámi in PDF work
  • Get more sma, smj texts to improve language recognition
    • Will get smj text today, 25th.
  • set up Tomcat for use with eXist and the propnouns db on the G5
  • fix bugs!

Maaren

  • download and install latest Marratech

Saara

  • Create a parallel corpora of the new testaments
  • add more texts to the graphical corpus interface
  • Implement parallel corpus upload in web upload script
  • remove headers and footers from pdf documents
    • Tomi did his part. There were some drawbacks, so the tool is not yet ready.
  • Implement server of the analysis tools.
    • Parallel processing implemented. Not otherwise finalized.
  • generate parallel corpus files manually (with Trond)
    • Started, but waiting for pdf-conversion.
  • Improve text_cat
    • The code is ready. I'll generate better language models for some languages before final testing.
  • fix bugs!

Sjur

  • fst gymnastics to add hyphenation and word boundary marks to hyphenation transducer
    • gymnastics done earlier - now a perl script is underway that will clean the output from overgeneration
  • name lexicon:
    • implement editing functions
    • finalise refactoring for multiple collections
      • continued to work on the specifications
    • implement improvements decided upon in Tromsø
  • review user and admin documentation for corpus access
  • write user account form, probably ask for copy of existing ones from the IT centre (with Trond)
  • start hiring process of linguist and programmer
  • help Børre finish i18n work of Forrest with a language override menu
    • almost DONE! It is working, only i18n of the language menu left, and sending in patches to Forrest, to make the updates part of the distribution
  • consider the problems of lexicalised derivations schewing analyses of derivation patterns
  • install eXist and our local copy of risten.no and propnouns on the G5
  • speller follow-up from the Tromsø meeting
  • discuss with Bård Eriksen about collecting smj texts (with Børre)
  • get instructions on how to use Marratech, and test it
  • fix bugs!

Thomas

  • sme G3 issue
    • this is still fixed, but not in the intended way
  • bug-fixing!
    • yeah!
  • refine smj proper noun lexica, cf. the propernoun-smj-lex.txt
    • not done
  • review user documentation for corpus access
    • not done
  • find and study all derived verbs in our corpus (depends on Trond)
    • not done
  • suggest which derivations could be generated (depends on Trond)
    • not done

Tomi

  • new proper name lexicon
    • data synchronisation of proper nouns between risten.no and CVS
    • XQuery refactoring and code development for our proper noun editor
    • new version of xml2lexc (based on ccat), should handle complex names correct: construct entries like we have now from the different parts of a complex name entry
    • implement improvements decided upon in Tromsø
  • read aligner docu, install, provide feedback
  • Set up the mechanism for the hash-mark transducer package
  • test the new xml output of the xml-tagged analyses
  • export corpus tools to /opt (with cron)
  • make speller and hyphenator make targets using M4
  • help Saara with JPedal
  • fix bugs!
  • other tasks:
    • worked on JPedal, to help Saara fix PDF conversion

Trond

  • better smj NT text
  • get fin, swe, nob and nno NT and OT in paratext format
  • fst gymnastics to add hyphenation and word boundary marks to hyphenation transducer
    • Worked on this with Sjur, it turned out to be quite hard. Will discuss it with experts.
  • make shell script wrappers for the most common commands for user friendlyness
    • This issue was passed on to the programmers.
  • write user account form, probably ask for copy of existing ones from the IT centre (with Sjur)
  • write documentation for our bound users, with pointers to the ordinary documentation
  • refine smj proper noun lexica, cf. the propernoun-smj-lex.txt
  • Get more sma, smj texts to improve language recognition
  • study corpus for language recognition errors, as well as paragraphs with mixed content
    • Done some work here, with Ilona.
  • generate parallel corpus files manually (with Saara)
    • Not done, but the aligner is now available in a debugged and faster version. Missing now is more parallel texts, and I have spent some time finding nob texts for our sme texts, with limited success so far.
  • block out the CG rule(s) that remove(s) the Der readings using M4
    • Also this issue has been passed on to the programmers, as the pseudocode is already written.
  • fix bugs!.

3. Documentation

TODO:

  • finish i18n work by adding a list of available language versions to each document (Børre with help from Sjur)
    • Sjur and Børre finished most late last Friday night (stopped around midnight) it is now working, and the patches needs to be sent to Forrest
  • make pdf set-up work on victorio (Børre)
    • working as it should on Victorio.
  • Write both user and admin documentation (Børre, review: Sjur, Thomas)
    • User documentation probably in several languages. This covers how to apply for an account, on what grounds one can apply, and pointers to documentation telling how to use the corpus.
    • Admin documentation, telling how we set the permissions to the corpus files, and whatever other processes and tasks needed to set up a corpus account.
  • add the new Words section to the site

4. Corpus gathering

Børre contacted several authors:

  • Jovnna Ánde Vest
  • Stig Gælok
  • Aage Solbakk

Børre will meet Stig Gælok today, he has a lot of texts in Lule Sámi.

Bård Eriksen was concerned that it would be too much work for them to deliver texts to us. Børre has asked for their book catalog, to be able to contact the authors directly.

TODO:

  • contact NSI (Børre)
    • not yet
  • contact authors (Børre, eventually Lene)
    • done, see above; no discussions with Lene
  • evaluate an agreement with Bård Eriksen helping us collecting smj texts (Børre and Sjur)
    • discussed with him

5. Corpus infrastructure

General

Our way of dealing with the conversion of input documents has now reached an advanced level. At some point we might consider to publish our results, to the benefit of the rest of the research community.

JPedal work: Tomi went through the source code and added an option that defines where the result goes. Didn't solve other issue with tagging.

TODO:

  • remove headers and footers in the PDF conversion (Saara)
    • still needs some work
  • Go through the java issues of JPedal (Saara, Tomi)
    • isn't quite delivering what we hoped, will need more work

User accounts and access

For details, see a previous meeting memo, as well as the memo from a dedicated meeting.

Shell access

TODO:

  • export to /opt (with cron) tools that the project team members find in their cvs tree (the bound users do not have a cvs tree, and therefore need these tools in /opt in order to conduct linguistic analyses) (Tomi)
    • Decision:
      • compiled transducers to /opt also in the future
      • scripts etc to /usr/local/share/bin/
  • make shell script wrappers for the most common commands for user friendliness (we must think of what commands they are) (Trond)
    • (first version of first script, teaksta.sh, was checked in, but it is still not working (the problem is a simple handling of input-output, some shell script literates should have a look at it)
  • write user account form, probably ask for copy of existing ones from the IT centre (Trond and Sjur)
  • possibly deploy the user account form as an HTML form (Børre)
  • write documentation for our bound users, with pointers to the ordinary documentation (Børre, Trond)
  • write documentation for how to apply for a user account (where's the form, to whom do I send the form, who needs it, etc.) (Børre)
  • make our own guidelines for the user application processing (Børre)
  • make a test user (Børre)
  • test corpus access as test user (Trond)

Web browser access

Has been discussed with Oslo. They will release a new version of the web interface in a couple of weeks. Further discussions delayed till then.

More texts to the graphical corpus interface:

TODO:

  • add text to the server (Lars)

Aligner

There has been a bug in the Bergen aligner, we will get a new (graphical) version shortly, and wait for that. When it arrives, we will do some conversion, still waiting for the command-line version, though. The second obstacle is the paucity of nob and fin text to parallel the sme ones.

TODO:

  • use the present aligner to generate some initial input for Oslo to test. ( Trond and Saara)
  • gather parallel texts (Trond)

Language recognition

New .wm files heve been made, with better performance. Saara, Ilona and Trond have been testing and refining the software. There still is some room for improvement. We now have a limit of 0 characters for paragraphs.

TODO:

  • Get more text of the poorly covered languages: sma, smj ( Trond, Børre)
    • sma: get the Bible texts (Trond)
  • study the mistakes our recogniser makes today (Trond, Ilona)
  • what about paragraphs with mixed content? Build a corpus of such paragraphs ( Trond)

6. Infrastructure

Xerox tools wrapped as servers

Feature request:

  • option for XML output from server

TODO:

  • improve and finish the present prototype (Saara)
    • done some, still more work to do

Hyphenator

Sjur got help from Saara to sketch a Perl solution to the overgeneration problem and a clean-up script is in the works. Will be ready this week.

TODO:

  • finish the hyphenator clean-up script (Sjur)
  • Update the sma hyphenator rule set with the insights gained from smj updates ( Trond during weekends)

Automatic Bugzilla reminder for untouched bugs

TODO:

  • give mail reminders a second try; ask Thor-Øivind for help if needed ( Børre)
    • At last I found a solution. Will implement it today!

M4

TODO:

  • make speller make targets that utilise M4 to produce normative and hyphenation transducers; also disamb variants (see next) (Tomi)

7. Linguistics

Derivation and spellers like Aspell

  • revert the CG rule that preferres lexicalised forms over derivations with M4 ( Trond wrote the M4 pseudocode, M4-literates to translate). In the beginning of sme-dis.rle there is an explanation of the pseudocode. Just search for the rules as explained there.
  • find and study all derived verbs in our corpus (Thomas)
  • suggest which derivations could be generated (Thomas)
  • lexicalise the rest (Thomas)

Semantic double-tagging of names

Waiting for the name conversion to take place before the disamb rules can be written. Further discussion delayed till then.

North Sámi

Nothing this week?

Lule Sámi

TODO:

  • refine smj proper noun lexica, cf. the propernoun-smj-lex.txt ( Thomas, Trond)
  • Schedule a T-T meeting this week - Wednesday.

8. Name lexicon infrastructure

Decided in Tromsø:

  • add logging facilities to the interface
  • add option to download local copies of the lexicon files directly from the db
  • batch editing (change all entries in the found set), should later be enhanced to allow selection of exceptions (the found set minus deselected items)
  • tag for excluding/including a name from certain applications
  • future epxansion: choose what info to display in the single language browser
  • display existing language entries when adding a new language to a record
  • add editor to change single, existing entries

Details can be found in the meeting memo.

TODO:

  • finish refactoring for multiple collections in the search interfarce ( Sjur)
    • worked on a specification (in the new CVSROOT/words/ section)
  • develop the needed XQueries and UI (Sjur, Tomi)
  • data synchronisation between risten.no and the cvs repo (Tomi)
    • discussion started on eXist-list, nothing useful came up. We need to reformulate the question from our perspective, and bring it up again ( Sjur)
  • add eXist and the proper noun interface to the G5 using Tomcat ( Sjur and Børre)

9. Tromsø meeting follow-up

TODO:

  • speller development - see the meeting memo. Separate follow-up next week.
  • Lule Sámi linguist (Sjur)

Speller data generation

We need to convert our Xerox lexicons to the format required by Polderland, Aspell, etc. The basic architecture for the conversion was decided upon in Tromsø, but it now needs to be implemented.

TODO:

  • start to plan the implementation of the speller data conversion/generation ( Tomi)

10. Other

Bug fixing

64 open Divvun/Disamb bugs (two down!), and 25 risten.no bugs

Guess: 1/3 of the bugs are fixed already (?)

Meetings and Marratech

TODO:

  • download and install newest Marratech ( Maaren)
  • we need instructions on how to use it, and test it (Sjur)

Task lists as iCal entries

TODO:

  • update Maaren's and Saara's installations to r430284 (Børre)

11. Next meeting, closing

Next meeting 2.10.2006 at 9: 30.

Closed at 10: 10.

Appendix - task lists for the next week

Børre

iCal

  • corpus collection:
    • contact Ája (Kåfjord), talk to Lene
    • send contracts to Čálliid Lágádus
    • contact Richard Valkepää at NSI about older Min Áigi and Áššu files
  • Move norwegian documents in Min Áigi from sme to nob
  • corpus access:
    • possibly deploy the user account form as an HTML form
    • Write both user and admin documentation (Børre, review: Sjur, Thomas)
  • set up Bugzilla automatic reminders for open issues
  • finish Forrest i18n and Sámi in PDF work
  • Get more sma, smj texts to improve language recognition
  • set up Tomcat for use with eXist and the propnouns db on the G5
  • add the new Words section to the site
  • fix bugs!

Maaren

  • On sick leave
  • download and install latest Marratech

Saara

iCal

  • Create a parallel corpora of the new testaments
  • add more texts to the graphical corpus interface
  • Implement parallel corpus upload in web upload script
  • remove headers and footers from pdf documents
  • Implement server of the analysis tools.
  • generate parallel corpus files manually (with Trond)
  • Improve text_cat
  • fix bugs!

Sjur

iCal

  • finish the hyphenator clean-up script
  • name lexicon:
    • implement editing functions
    • finalise refactoring for multiple collections
    • implement improvements decided upon in Tromsø
  • review user and admin documentation for corpus access
  • write user account form, probably ask for copy of existing ones from the IT centre (with Trond)
  • start hiring process of linguist and programmer
  • finish i18n work of Forrest
  • consider the problems of lexicalised derivations schewing analyses of derivation patterns
  • install eXist and our local copy of risten.no and propnouns on the G5
  • speller follo-up from the Tromsø meeting
  • get instructions on how to use Marratech, and test it
  • fix bugs!

Thomas

iCal

  • work with Polderland phonetic rules
  • bug-fixing!
  • refine smj proper noun lexica, cf. the propernoun-smj-lex.txt
  • review user documentation for corpus access
  • find and study all derived verbs in our corpus (depends on Trond)
  • suggest which derivations could be generated (depends on Trond)
  • meeting with Trond Wednesday on smj proper nouns

Tomi

iCal

  • new proper name lexicon
    • data synchronisation of proper nouns between risten.no and CVS
    • XQuery refactoring and code development for our proper noun editor
    • new version of xml2lexc (based on ccat), should handle complex names correct: construct entries like we have now from the different parts of a complex name entry
    • implement improvements decided upon in Tromsø
  • export corpus tools to /opt (with cron)
  • make speller make targets using M4
  • start to plan the implementation of the speller data conversion/generation
  • fix bugs!

Trond

iCal

  • better smj NT text
  • get fin, swe, nob and nno NT and OT in paratext format
  • fst gymnastics to add hyphenation and word boundary marks to hyphenation transducer
  • make shell script wrappers for the most common commands for user friendlyness
  • write user account form, probably ask for copy of existing ones from the IT centre (with Sjur)
  • write documentation for our bound users, with pointers to the ordinary documentation
  • refine smj proper noun lexica, cf. the propernoun-smj-lex.txt
  • Get more sma, smj texts to improve language recognition
  • study corpus for language recognition errors, as well as paragraphs with mixed content
  • generate parallel corpus files manually (with Saara)
  • block out the CG rule(s) that remove(s) the Der readings using M4
  • meeting with Thomas Wednesday on smj proper nouns
  • fix bugs!.