Closed pagelj closed 6 years ago
My plan was to export the annotations from XMI into TEI, and then to read them in into the pipeline. The exporting to TEI has not happened yet, though.
XMI -> TEI export is now addressed in https://github.com/quadrama/DramaNLP/issues/46
Mentions: <rs ref="">Sara</rs>
Annotated disagreements: <rs ref="e1 e2" func="or">bla</rs>
Mentionend.figure_id
column in CoNLL styleAnother possible value is func="and"
, which is the default and can also be achieved by omitting func
Aggregated mentions, i.e. mentions that point to more than one entity are simply treated as two separate entities with the same span, e.g. they,(CATO)|(ARSENE),(cato)|(arsene)
in the Mentioned.figure_surface
and Mentioned.figure_id
column
Integrate the .xmi annotations of coreference into the pipeline.
mention_id
andmention_surface
to every token of a mention (right now, it's only the first token of a mention) (https://github.com/quadrama/DramaNLP/commit/234c88531087c4ea99ff9d3e728297bf2c3803b6)CastFigure
(https://github.com/quadrama/DramaNLP/commit/ebb6f427ff07af9bc91c212e352d00574f33c4c3)