Meeting_2006-10-16

• Date: 16.10.2006
• Time: 09.30 Norw. time
• Place: Where we are
• Tools: SubEthaEdit, iChat

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

Opened at 10: 11.

Present: Børre, Maaren, Sjur, Thomas, Tomi, Trond

Absent: Saara

Agenda accepted as is.

Børre

• corpus collection:
  • contact Ája (Kåfjord)
    • Not done
  • discuss access to older Min Áigi and Áššu files with Richard Valkeapää
    • I will provide him with a renaming script, and then we will get the Áššu and Min Áigi files.
• Move norwegian documents in Min Áigi from sme to nob
  • Not done
• set up Bugzilla automatic reminders for open issues
  • Not done
• finish Forrest i18n work (pdf)
  • Not done
• set up Tomcat for use with eXist and the propnouns db on the G5
  • Not done
• investigate use of proxies for Maaren, for her to be able to use iChat AV
  • Done. SquidMan is a possible solution, will test as soon as possible
• fix bugs!
  • Not done
• Other
  • Worked a lot with the aligner, tca2.

Maaren

• download and install latest Marratech
  • with the help of SquidMan and http proxy, this is now obsolote - good bye, Marratech!

Saara

• add more texts to the graphical corpus interface
• finalize server of the Xerox tools.
• generate parallel corpus files manually (with Trond)
• Improve text_cat
• export corpus tools to location available to all (with cron), cf news disc.
• Improve hyph-filter.pl
• Implement M4-rules for sme-dis.rle
• improve smj LM based on the new, larger corpus
• help Trond with some shell commands
• fix bugs!

Sjur

• name lexicon:
  • refactor SD-terms editor code
    • started
  • implement missing propnouns editing functions
    • nothing
  • implement improvements decided upon in Tromsø
    • nothing
• move corpus user doc issue to Bugzilla
  • nope
• hire linguist and programmer
  • nothing last week
• finish i18n work of Forrest
  • nothing
• install eXist and our local copy of risten.no and propnouns on the G5
  • nope
• add normative M4 macros to the twol code; utilize it in the make file
  • done
• fix bugs!
  • fixed one with Thomas and Trond
• other things done:
  • spent a lot of time improving hyphenation and speller generation for Polderland, partly together with Thomas and Trond. Much has improved, but there are still issues, and hyphenation for smj does not yet compile

Thomas

• work with Polderland phonetic rules for smj
  • done
• refine smj proper noun lexica, cf. the propernoun-smj-lex.txt
  • nothing this week
• find and study all derived verbs in our corpus (depends on Saara)
  • have not started yet
• suggest which derivations could be generated (depends on Saara)
  • not yet started

Tomi

• continue implementation of the speller lexicon conversion
  • continued
• fix bugs!
  • fixed some bug

Trond

• Discuss shell script wrapper teaksta.sh with Saara
  • Not done
• write user account form, probably ask for copy of existing ones from the IT centre (with Sjur)
  • Not done
• write documentation for our bound users, with pointers to the ordinary documentation
  • Not done
• refine smj proper noun lexica, cf. the propernoun-smj-lex.txt
  • Not done
• Get more sma texts to improve language recognition
  • Not done
• study paragraphs with mixed content
  • Not done
• add corpus artice framework to Bugzilla
  • Not done
• add corpus user accounts and access issues to Bugzilla
  • Not done
• fix bugs!.
• As can be seen, this week has been devoted to totally different things, first and foremost to getting our first analysed texts ready for corpus inclusion. (and to hyphenation...) I'll just have to move the tasks to next week.

TODO:

• finish i18n work (Børre and Sjur)
  • set up Tomcat at the faculty server, and install Forrest as war. Needs testing before being deployed on the server! Thus, install it on the G5 first and see that it works, then on the faculty server.
  • i18n does not work in PDF ("Table of Content" won't translate)
    • still doesn't work
• 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.
    • move these issues into Bugzilla (Sjur)

Børre talked to Richard Valkeapää, they have several computers, and have the data available. But they have problems with their backup system and interaction between file names on Windows and MacOS X. Børre will help them, and we will receive our data.

TODO:

• discuss corpus transfer with NSI (Børre)
• continue to contact authors (Børre)

User accounts and access

The whole issue should be moved into Bugzilla as several tasks.

TODO:

• add the issue with subissues to Bugzilla (Trond)

More texts to the graphical corpus interface:

Trond, Børre and Saara aligned and sent the first parallell corpus files to Oslo. One milestone further : -)

TODO:

• align texts, analyse, and send to Lars (Trond, Saara)
• add text to the server (Lars)

Aligner

Børre has made some important improvements to the Bergen Aligner. The Aligner compiles better and is now able to save intermediate result files (4 out of 5 files, that is, so there is still some work to do), in order to control memory usage. In this way, we are hopefully able to contribute to the Bergen development.

TODO:

• gather parallel texts (Trond)
  • Trond is working on it
• try out NT alignment strategies (Saara)

Language recognition

TODO:

• get more sma texts, first the Bible / NT (Trond)
• improve the smj LM (Saara)
  • done
• what about paragraphs with mixed content? Build a corpus of such paragraphs ( Trond)

Xerox tools wrapped as servers

The Divvun project needs generation pretty soon, and Tomi can add generation to the server if he needs it. Possible conflicts in the code will have to be resolved when Saara is back.

TODO:

• improve and finish the present prototype (Saara)
• add generator (Tomi)

Hyphenator

Sjur has hyphenated the pl-wordlist.txt file (the full-form word list generated for Polderland), and now has about 10Gb of hyphenated data on his computer! Forgot to use time, but it took somewhere between 36 and 48 hours! There's a lot of noise and problematic issues in the data, but a cleaned version will be sent off to PL for their perusal. The cleaned version is 651Mb, just slightly larger than the original input file, which was 627Mb.

Hyphenation for smj still does not work, as the hyphenation rules file does not compile. We have received valuable feedback from Lauri Karttunen, though, as we reported a UTF-8 bug in the latest versions of the Xerox tools.

Overgeneration problem:

Abakalikilaččabuččaideamet      A^ba^ka^li^kilač^ča^buč^čai^dea^met
Abakalikilaččabuččaideamet      A^ba^ka^li^kilač^ča^buč^čai^dea^met
Abakalikilaččabuččaideamet      A^ba^ka^li^kilep^po^žiid^dá^met
Abakalikilaččabuččaideamet      A^ba^ka^li^kilep^po^žiid^dá^met
Abakalikilaččabuččaideamet      A^ba^ka^li^kilab^bo^žiid^dá^met
Abakalikilaččabuččaideamet      A^ba^ka^li^kilab^bo^žiid^dá^met
Abakalikilaččabuččaideamet      A^ba^ka^li^ki#lač^ča^buč^čai^dea^met
Abakalikilaččabuččaideamet      A^ba^ka^li^ki#lep^po^žiid^dá^met
Abakalikilaččabuččaideamet      A^ba^ka^li^ki#lep^po^žiid^dá^met
Abakalikilaččabuččaideamet      A^ba^ka^li^ki#lab^bo^žiid^dá^met
Abakalikilaččabuččaideamet      A^ba^ka^li^ki#lab^bo^žiid^dá^met

Unrecognised input:

abbalaččaboiid          abbalaččaboiid          +?
Abbalaččaboiidda        Abbalaččaboiidda        +?
abbalaččaboiidda        abbalaččaboiidda        +?
Abbalaččaboiiddáde      Abbalaččaboiiddáde      +?
abbalaččaboiiddáde      abbalaččaboiiddáde      +?
Abbalaččaboiiddádeguin  Abbalaččaboiiddádeguin  +?

What is the percentage of unrecognized input? (i.e. of word forms generated by the generator but not accepted by the analyser)? Answer:

• Wordlist generated = 27 354 884
• unrecognised wordforms = 11 284 570
• percentage unrecognised = 41,3 %

This is way too high, and must be investigated. The input is generated from our normative, non-circular version of the lexicon, and should thus only contain recognised word forms already in the lexicon.

TODO:

• Update the sma hyphenator rule set with the insights gained from smj updates (Trond during weekends)
• linguistic issues in the hyphenator: ( Sjur using M4)
  • shortening in compounds (grossly overgenerates in the hyphenator output)
    • done - the shortening rules are now filtered out by M4 when compiling the twol for hyphenation and other normative uses
• consider case conversion problems (Saara)
• check #- when comparing input string with hyphenated string - the hyphen could be part of the input string, keep it if it is, discard the hyphenated string if not (Saara)
• investigate unrecognised input word forms (Maaren, Thomas, Trond, Sjur)

Automatic Bugzilla reminder for untouched bugs

Some perl-libraries needed by Bugzilla weren't in the path, causing it to not work. Adding them should fix the issue.

TODO:

• fix the remaining issues (Børre)

M4

TODO:

• make speller make targets that utilise M4 to produce normative and hyphenation transducers - postponed a while (Tomi or Sjur)
  • done
• disamb variants for derivations (see next) (Saara)
  • include make as part of a shell script, which will run the M4 processor on the disamb rules (Saara)
    • done - but the disambiguation REMOVE rules need to be added in again for the benefit of the disamb project. What the syntacticians really want is to have this done in the cg-3 interface.

Names and multilinguality

We need a more principled approach to this.

Background: the name lexicon is getting attention from the SD name/terminology sections, and they would like to use our name lexicon also for public searching.

Observations:

1) Multilinguality is always optional.

2) We can observe that "foreign" names in texts follows a domination pattern: majority language forms can be found in minority language texts as real names ("Kautokeino produkter"), whereas minority language names almost always occur in majority language texts as citations. And citations should not be considered a natural part of the text.

3) When looking at our name classification, multilinguality varies according to:

Ani - weak/none? (pet, myth anim.  names)
Fem - weak (informative)
Mal - weak (informative)
Obj - strong
Org - strong
Plc - strong for the national and country names, weak (informative) for foreign
       names
Sur - none
Tit - strong (titles)

Suggestion:

We need to reconsider the all names in all languages policy. That policy is valid only for Fem, Mal, and Sur (and Ani and Tit?). For Obj, Org, Plc the rule should be that if they have multilingual names, each name should only be used in it's own language. Then we need a modification saying that majority language names can be included in minority language lexicons if attested in our corpus. Also, the majority language varies according to country (obviously), which means that in a speller context, we might consider tailoring spellers for each country, leaving out noise relating to majority language names from another country.

A further issue is whether we should reconsider our cohort policy. Today, Sur and Plc are different readings. An alternative would be to have them as secondary tags, not in conflict with each other:

"<Trosterud>"
        "Trosterud" N Prop Sur Sg Nom <<< @HNOUN
        "Trosterud" N Prop Plc Sg Nom <<< @HNOUN
"<Trosterud>"
        "Trosterud" N Prop Sg Nom <Sur> <Plc> <<< @HNOUN
"<Trosterud>"
        "Trosterud" N Prop Sg Nom &Sur &Plc <<< @HNOUN

Derivation and spellers like Aspell

• enclose the CG rule that preferres lexicalised forms over derivations in M4 ( Saara)
  • done
• find and study all derived verbs in our corpus (Thomas)
• suggest which derivations could be generated (Thomas)
• lexicalise the rest (Thomas)

North Sámi

We have a lot of overgeneration, mainly possible derivations that are spelled out. We need to investigate what is going on, and evaluate whether it is a problem.

TODO:

• investigate the generated word form list sent to Polderland - use the command make pl-wordlist TARGET=sme in victorio ( Maaren, Thomas)

Lule Sámi

TODO:

• refine smj proper noun lexica, cf. the propernoun-smj-lex.txt ( Thomas, Trond)
• hire new linguist (Sjur)

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:

• develop the needed XQueries and UI (Sjur, Tomi)
• turn Tomcat on on the G5; send admin username and password to Sjur ( Børre)
• add eXist and the proper noun interface to the G5 (Sjur)

Postponed:

• data synchronisation between risten.no and the cvs repo
• 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

News

The Greenlandic speller was released today. It is available at the website of the greenlandic language council. It is available for free, and is based on the same technology as our own.

Speller data generation

It reads the lexc files, reads each word, and requests (or will request) the generator server to expand each base form to its full paradigm.

Still missing: the generator isn't yet available, Tomi should implement a first version while Saara is away. It should allow requests for:

• single word forms
• reduced paradigm
• full paradigm (all possessives, clitics etc, including compound forms)

The Java code needs to be checked in into cvs, suggested location:

gt/src/lexc2xspell

TODO:

• add code to cvs (Tomi)
• implement generator server based on Saara's code (Tomi)
• specify compounding behaviour info for the lexicon (Thomas, Trond, Sjur)
• add hyphenation points to the generated output (Tomi)

Bug fixing

66 open Divvun/Disamb bugs, and 24 risten.no bugs

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

Meetings and the SD Firewall

TODO:

• investigate use of proxies for Maaren ( Børre)
  • done and solved nicely!
• document SquidMan use at SD (Børre)

Task lists as iCal entries

TODO:

• update Maaren's installation to r430284 (Børre)

Words section

You all need to check out CVS/words, and link to the relevant place, cf how gt/doc/ is linked.

Next meeting 23.10.2006 at 9: 30.

Sjur will be away on Thursday and Friday this week. Trond will be away next week.

Closed at 11: 16.

Boerre

• corpus collection:
  • contact Ája (Kåfjord)
  • discuss access to older Min Áigi and Áššu files with Richard Valkeapää
• Move norwegian documents in Min Áigi from sme to nob
• set up Bugzilla automatic reminders for open issues
• finish Forrest i18n work (pdf)
• set up Tomcat on the G5 for use with eXist and the propnouns db, as well as Forrest in its i18n form
• document SquidMan use at SD
• fix bugs!

Maaren

• investigate the generated word form list sent to Polderland - use the command make pl-wordlist TARGET=sme in victorio
• investigate unrecognised input word forms in the hyphenator

Saara

• add more texts to the graphical corpus interface
• finalize server of the Xerox tools.
• generate parallel corpus files manually (with Trond)
• Improve text_cat
• export corpus tools to location available to all (with cron), cf news disc.
• Improve hyph-filter.pl
• help Trond with some shell commands
• fix bugs!

Sjur

• name lexicon:
  • refactor SD-terms editor code
  • implement missing propnouns editing functions
  • implement improvements decided upon in Tromsø
• move corpus user doc issue to Bugzilla
• hire linguist and programmer
• finish i18n work of Forrest
• install eXist and our local copy of risten.no and propnouns on the G5
• investigate unrecognised input word forms in the hyphenator
• decide how to specify compounding behaviour info in the lexicon
• fix bugs!

Thomas

• refine smj proper noun lexica, cf. the propernoun-smj-lex.txt
• find and study all derived verbs in our corpus
• suggest which derivations could be generated
• investigate unrecognised input word forms in hyphenator
• investigate the generated word form list sent to Polderland - use the command make pl-wordlist TARGET=sme in victorio
• decide how to specify compounding behaviour info in the lexicon

Tomi

• continue implementation of the speller lexicon conversion
• make generator as server, based on Saara's code
• add lexc2xspell code to cvs
• add hyphenation points to the generated output
• fix bugs!

Trond

• refine smj proper noun lexica, cf. the propernoun-smj-lex.txt
• Get more sma texts to improve language recognition
• study paragraphs with mixed content
• add corpus user accounts and access issues to Bugzilla
• investigate unrecognised input word forms in the hyphenator
• decide how to specify compounding behaviour info in the lexicon
• fix bugs!.