Statistics from ACE05 evaluation plan(dead link - archived version)

From Durrett & Klein (2014)[1]: "ACE 2005 corpus (NIST, 2005): this corpus annotates mentions complete with coreference, semantic types (per mention), and entity links (also per mention) later added by Bentivogli et al. (2010). [...] train/test split from Stoyanov et al. (2009), Haghighi and Klein (2010), and Bansal and Klein (2012)."

Versions Edit

  • Official contest: test data is only available to participants
  • LDC2005E18: enlarged version
  • LDC2006T06: further enlarged version
  • Stoyanov et al.'s split[2]: newswire only, documents unclear from which set, 57 documents for training, 24 for testing (ratio 70/30)
  • Rahman and Ng's split[3]: full 599 documents from LDC2006T06, split into 482 documents for training, 117 testing (ratio 80/20), balanced between genres

Statistics from ACE05 evaluation plan(dead link - archived version)

See also Edit

References Edit

  1. Durrett, G., & Klein, D. (2014). A Joint Model for Entity Analysis: Coreference, Typing, and Linking. In Transactions of the Association for Computational Linguistics (Vol. 2, pp. 477–490). Retrieved from
  2. V Stoyanov, N Gilbert, C Cardie, and E Riloff. 2009. Conundrums in Noun Phrase Coreference Resolution: Making Sense of the State-of-the-art. In Associate of Computational Linguistics (ACL).
  3. A Rahman and V Ng. 2009. Supervised models for coreference resolution. In Proceedings of the 2009 Conference on Empirical Conference in Natural Lan- guage Processing.