Problem running the create_statistics.sh script about parzu

[rsennrich]

thanks; this was a bit more involved, since the position of features, and the formatting of the morphological features, is also different. The script should now support both current versions of TüBa-D/Z and HDT.

[johann-petrak]

Thanks!

When I try this, it now looks as if the create_statistics.sh script can directly digest the original CONLL-format version of the HDT while what I did was to use the statistics/stats_creator/hamburg_treebank_to_conll.py script first to convert the HDT to Conll, and that file then created the problem with the incorrect number of fields.

However, when I tried now to run the create_statistics.sh script with the output of the hamburg to conll script, I got an error

ERROR: '<meta-call>'/1: Undefined procedure: w/7
   Exception: (11) w(_2524, _2530, _2558, ne, _2562, _2536, _2542) ? 

This is not really a problem since using the original HDT conll file is probably better anyways, but it may be confusing because the README.md file still suggests that this should work, if I understand correctly.

[rsennrich]

thanks for pointing that out; I removed the hamburg_treebank_to_conll.py and updated the README - taking the CONLL-formatted HDT is now the supported way to use it.