Closed ghost closed 5 years ago
So is <root>/dev-xml/UCCA_English-Wiki/705008.xml
. There're only two sentences that can't be converted among all UCCA English Wiki data.
We fixed these two issues after the shared task: https://github.com/UniversalConceptualCognitiveAnnotation/UCCA_English-Wiki/commit/ed13dee6b4a089dc5755da07f6085902f6c0b4ae#diff-05d5cce54c47b675f644021ad0adc51a https://github.com/UniversalConceptualCognitiveAnnotation/UCCA_English-Wiki/commit/61560134ab292e77a7eb337a1952f3e4cd16d8d2#diff-05d5cce54c47b675f644021ad0adc51a
To avoid failures for invalid input passages, use --no-validate-oracle
: https://github.com/danielhers/tupa/blob/master/tupa/oracle.py#L63
Let me know if it works.
Deleting the 3 lines works. Thank you!
I tried to use my modified edition of the file
tupa/test_oracle.py
to convert UCCA input of this competition to chains of oracles. I was able to work on most inputs, but failed on the file<root>/dev-xml/UCCA_English-Wiki/705006.xml
in their "public" dataset.The output message shows: