search

otakar-smrz / elixir-fm

ElixirFM Functional Arabic Morphology
http://elixir-fm.sf.net/
42 stars 12 forks source link

POS tagger existence #5

Open aosaimy opened 7 years ago

aosaimy commented 7 years ago

Hi,

Are you guys aware of any POS tagger that uses Elixir and selects most proper analysis from the context?

Prague Arabic Treebank is available and uses the same POS tagset of Elixir. This should ease the build a POS tagger trained on PADT.

otakar-smrz commented 7 years ago

I have not yet shared the relevant PADT code in this repository. The AQMAR extension to TrEd would combine ElixirFM annotations with the output of the MADA tagger.

https://github.com/otakar-smrz/elixir-fm/blob/master/README.md https://github.com/otakar-smrz/elixir-fm/tree/master/TrEd

The NLP technology is leaping forward, I would expect some tagger to be successfully trained on a reasonably small corpus of seed annotations. Please keep me in the loop of your explorations!