PMC proposals/Stable version of apertium-sh-sl

From Apertium
Jump to navigation Jump to search


As part of FP7 project Abu-MaTran, it is proposed to have sh–sl (or hbs–slv if the other initiative presented to the PMC today is passed) improved.

The language pair seems to work OK in the sh→sl sense but not so well in sl→sh (apparently it has not been testvocked).

Improving this language pair would be nice for the first milestone of the project Abu-MaTran (June 30) as Croatia will be joining the EU on July 1. It would be used to pivot much in the same way as described in Antonio Toral's EAMT 2012 paper, using en→sl (SMT) + sl→hr (Apertium). However, project Abu-Matran does not hire anyone before 2014, so there is no available workforce. Also, we cannot schedule it as a GSoC task, or, in any case, it should be scheduled as a very early part of a GSoC task that could then continue improving it.

The task would consist of "testvocking" the language pair to avoid @'s and #'s in both directions and to make it reasonably usable in both directions, with the minimum possible amount of creation of linguistic data.

The task should be finished by April 30 and whoever takes it on will get a reward of €500 from Apertium Google Summer of Code funds.

Proposed by: User:Mlforcada Seconded by: Francis Tyers

In detail



  • Comment: From what I saw there is not a lot of work to do to finish the testvoc, however, the coverage is still fairly low (around 79-80%), thus I think it would be good to include expansion of dictionaries --- by (semi-) automatic methods. - Francis Tyers 14:47, 18 February 2013 (UTC)
  • Comment: Hrvoje Peradin and Filip Petkovski have offered to do the work, splitting the reward on a 25%/75% basis. - Francis Tyers 17:26, 18 February 2013 (UTC)



  1. Jacob Nordfalk 16:56, 17 February 2013 (UTC)
  2. Juan Antonio Pérez 08:30, 18 February 2013 (UTC)
  3. Francis Tyers 14:43, 18 February 2013 (UTC)
  4. Jimregan 18:17, 18 February 2013 (UTC) (Specifically for Hrvoje and Filip being funded, though I see no objection to running it as a bounty).