Adding extended nlp_uk morphological tagging for PressMint-UA

[Dandelliony]

Dear PressMint Team,

We are considering adding morphological annotations generated by the nlp_uk tagger to the annotated files. This tagset is our standard for Ukrainian (https://github.com/brown-uk/dict_uk/blob/master/doc/tags.txt).

If I’m not mistaken, you have a similar option for MULTEXT-East annotation and section 6.1.1 of the documentation (https://clarin-eric.github.io/PressMint/#sec-ana-words) also allows this.

The only issue is that we use colons to separate tags. You can see an example of the annotation here: https://github.com/Dandelliony/PressMint/blob/data/Samples/PressMint-UA/Sources/Rada_1910_245_Aresht_semyi.tagged.xml

Please let us know if there are any other points to consider; we are currently reviewing the documentation on this topic.

Best regards,
Arsenij

[TomazErjavec]

Yes, you are right and can treat it like the MULTEXT-East tagset, except that you change the colon to some character that does not appear in the tags and is also legal in an XML ID value and then define somewhere the features structures with the appropriate IDs, just like in the documentation (cf. also https://nl.ijs.si/ME/V6/msd/tables/msd-fslib2-sl.xml)

However, given that your tagset is in fact already composed of features, maybe you could endcode them directly, so that you would have for e.g. <token value="МОГИЛІВ" lemma="Могилів" tags="noun:inanim:m:v_naz:prop:geo" /> something like <w lemma="Могилів" ana="uk:noun uk:inanim uk:m uk:v_naz uk:prop uk:geo">МОГИЛІВ</w>.

Then your prefix defintition would define the uk prefix to point to a file which contains <fLib> and this one features like <f xml:id="noun" name="CATEGORY"><symbol value="noun"/></f>.

Either would be fine I think, except if @matyaskopp has some further insights!