First import from private/Ingestion
Works for tree_tagger, marmot, opennlp, and corenlp foundries.
Current limitations:
1. conllu2korapxml is still missing
2. No tests so far. They should probably be added when conllu2korapxml
is available, using conllu2korapxml|korapxml2conllu-cycles.
3. Only morpho-syntactic annotation layers are handled, no dependencies
or constituencies.
4. Instead of starting from base we currently start from the
annotations, so that when annotations are missing in some places no data
at all will be printed.
5. For sentence segmentations, the morpho-synractic annotations are used
instead of the base sentence structure.
Change-Id: Id84133a1637c9c7c7f0235f57827a052d78882b3
3 files changed