Meeting rus_oahpa 5.9. 2013


Present: ''Robert, Heli, Laura, Tore, Trond''

!!!Agenda


* Status quo
* Leksa
* Bugzilla
* Next meeting

!!!Status quo


!!Leksa status

Leksa is online, in one direction, at [http://testing.oahpa.no/rusoahpa/leksa]

More translations can be added at lib.

Turning direction: Cip has been added Heli some scripts, that she is
improving, so a testing for X2rus is forthcoming.

Trond has added Norwegian (and a couple of English) synonyms.

$GTHOME/ped/inc/MiP

Add synomyms?


!!Database generation

A list of words not being generated have been generated.
Heli gave them to Robert.

The database has been reinstalled. POS changes since last meeting 
are thus implemented, the Adv-as-N issue is thus resolved.

!! Linguistics improvement

Robert has worked on typos and other issues in the apertium fst,
with data from Heli. We will discuss with Francis later on.

The automated way to generate everything.

Testing procedures

* expand the whole lexicon and search everything
* make a list of lemma + POS and send to the generator

Robert and Trond have been looking at that.

!! One level vs. two levels

Lexc
Meeting


!!!Leksa

Sveta was excited, but wanted us to turn it around.


!!Leksa X2rus


Time plan: Beginning of next week (optimistic).

Note: the same treatment of stress as for Morfa.

As soon as Leksa is up


Handling perfectivity in Leksa:

nob2rus should contain the following correct answers:

* читать/прочитать (stat="pref")
* читать
* прочитать
* прочитать/читать


!!!Apertium format vs. hfst format (one fst vs two fsts)

We consider moving to a twolc + lexc format.

PRO:

* Scalability
* Readability of the code

Laura to contact Francis for a discussion.

ped/rus/inc/MiPListAlpha_corrected.xml



!!!Bugzilla

* ''агенство / агенТство''
* ''суда - сюда''
* ''прав'' = ''ты прав''

30 words are missing, and can be added. They are not
recognised by the fst.

Heli will filter out whatever is not generated.
Errouneous form should thus be corrected, missing forms may


!!!Next meeting

Thursday 19. sept at 8.30

Topics including:

* Leksa X2rus
* fst: hfst vs. apertium format