Open Ansa211 opened 6 years ago
It simply is so - the Czech data does not have heads and deprels filled in :-( Should we erase any attribute that has a single value? That would mean that not all corpora would have the same attributes, but it would be less confusing for anyone trying to use an attribute that actually is not there.
Note: I already erased one attribute that only contained a few values in the Polish data and otherwise no value at all... it was column number 9 in the original .conllu file, I don't know what information it is supposed to carry.
This is also related to the question whether we should convert the corpora that come without any .conllu file at all.
https://lindat.mff.cuni.cz/services/kontext-staging/ansa/view?ctxattrs=word&attr_vmode=visible&pagesize=40&refs=%3Ddoc.id&q=~qoC6kvit&viewmode=kwic&attrs=lc%2Cfeats%2Chead%2Cdeprel%2Cmisc&corpname=parseme_cs_a&attr_allpos=kw&structs=&fromp=1