search

wikilinks / conll03_nel_eval

Python evaluation scripts for AIDA-formatted CoNLL data
Apache License 2.0
20 stars 4 forks source link

Add TAC reader #41

Closed benhachey closed 10 years ago

benhachey commented 10 years ago

Requires some refactoring in data. At least:

Data itself can remain agnostic about offset type (token vs char).

jnothman commented 10 years ago

Do we want to support TAC < 2014?

On 14 May 2014 10:51, Ben Hachey notifications@github.com wrote:

Requires some refactoring in data. At least:

  • Disentangle sentence annotations from mentions;
  • Add entity type to mention.

Data itself can remain agnostic about offset type (token vs char).

— Reply to this email directly or view it on GitHubhttps://github.com/wikilinks/conll03_nel_eval/issues/41 .

benhachey commented 10 years ago

I believe the format is the same. The difference in TAC14 is just that the systems produce the query xml file as well as the linking annotations tab file.

benhachey commented 10 years ago

Done in tac at 7ccff51200.