#17 Support for overlapping syntactic annotations

[Adam Radziszewski]

It would be useful if Poliqarp could handle corpora where syntactic annotation may be decompposed into several partially independent “channels” of annotation. For instance, one channel could contain NP chunks compliant with NKJP definition, the other could contain NP chunks coming from other source.

E.g.

NP-NKJP:  [some tokens] are for [the sake] [of an example]
NP-Other: [some tokens] are [for the sake   of an example]

My request is that Poliqarp is able to read such corpus first. Obviously, in queries one would have to specify which annotation is to be captured, NP-NKJP or NP-Other.

In other words, it would be very useful if the input format for Poliqarp could contain overlapping annotations, not limited to NKJP-style groups-within-other-groups.

This is important to be able to handle some corpora, e.g. KPWr (http://nlp.pwr.wroc.pl/pl/narzedzia-i-zasoby/kpwr)