160804

Samest meeting, 4th August 2016

Present: Fran, Heiki, Heli, Jaak, Trond, Tiina

Agenda

  • MT
  • Est FST
  • Oahpa
    • General
    • vro-oahpa
    • est-oahpa

MT

From last meeting: "Fix the system".

  • Tag harmonising
  • CG outsourcing
  • makefiles for all directions in apertium
    • fin-est, est-fin ok
    • sme-fin ok, fin-sme ?
  • transfer rules in .t1x and lexical selection in .lrx
  • CG: Finnish objects
  • Tags: Pron, Adp etc.
  • Compounding

FST

Summing up the summer:

  • Perhaps 90% of the FST tag pleonasms (Act vs. Act Ind etc) are gone
  • Some CG rules have been added, to disambiguate more
  • Summary on CG recall (removing correct entries):
    • The recall from a small corpus is 99%
    • The recall from the univ treebank is 90%. Half of the errors are due to tag differences, so a more realistic estimate would be a recall of 95%
    • A case in point: nuorille (N Pl All/A Pl All) (these are free rides)
    • participle vs adjective: innostamana (V AgPrc Sg Ess/A Sg Ess)
    • noun vs postposition: avulla (N Sg Ade/Po)

Apertium setup

The list "MT" above is good. Status is open, so let us have a physical meeting in Tromsø next thursday.

Keit Mõisavald attended RBMT summer school in Alacant. Worked on Fin-Est.

Est FST

Not much has happened during sun shining. Back on track soon.

Oahpa

General

We will start brainstorming next week, scheduling for mid-september week.

vro-oahpa

Sulev and Heli had a meeting on 30.6. in Tartu and set up a todo-list that should be done before the autumn term starts (or in October, latest). Most of the things are related to Morfa-C:

  • avoid all kind of repetitions
  • add more words to the semantic classes
  • add a possibility to read aloud the Morfa-C questions using the Võro speech synthesis (http://www.eki.ee/~indrek/voru/index.php)

est-oahpa

Summer holiday.

rus-oahpa

Tiina has been working on rus-oahpa during the summer, as part of the preparations for Oahpa 2.0.

Differences are mainly linguistic.

The challenge is to combine languages with different needs in the same infrastructure.

Next meeting

August 19th, 0900 Norwegian time, 1000 Estonian time

Things to do until then:

  • mt
    • fin.fst + fin.cg improvement
    • tag conversion gt to ap
    • preparing for tag changes fin - est + fin - sme
  • est.fst
    • recover from vacation
  • oahpa
    • vro - work on the todo-list
    • est - set up the next goals