[mrp-users] update to german syntactic companion trees

[Stephan Oepen]

dear colleagues:

one team has brought to our attention that no less than 417 of the
german DRG gold-standard graphs lacked a corresponding syntactic
companion parse.  we apologize for this omission!  for those of you
using our syntactic companion trees, we are now providing an update
for the file 'mrp/2020/cl/companion/deu.mrp' by direct download:

http://svn.nlpl.eu/mrp/2020/public/deu.mrp

further, we are still investigating some (seemingly) surprising anchor
values in the czech syntactic companion parses.  like for the german
UD trees above, we may want to issue an update to the czech companion
data one of the next few days.

the LDC has confirmed that the MRP 2020 evaluation data will be
available starting tomorrow, no later than 12:00 noon at the US east
coast (but most likely already earlier in the day).  to access the
parser inputs for the evaluation phase, please re-download the full
LDC2020E05 data package from the LDC server.  the latest version
contains new directories 'cf/evaluation/' and 'cl/evaluation/'

finally, we would like to remind you of a technical difference between
the english vs. chinese AMR graphs: the CAMR annotations include
gold-standard anchorings, i.e. formally the chinese AMR graphs
instantiate our Flavor (1) (formally, if not linguistically, much like
the EDS graphs), while english AMR exhibits what we call Flavor (2).
everything provided in the gold-standard annotations shall be
predicted by participating parsers and will be scored, including
anchorings in the chinese AMR graphs.

best wishes, oe