Catalan data from the AnCora corpus.
The original annotation was done in a constituency framework as a part of the AnCora project at the University of Barcelona. It was converted to dependencies and used in the CoNLL 2009 shared task. The CoNLL 2009 version was later converted to HamleDT and to Universal Dependencies.
The GNU license is inherited from the original dataset, downloaded from the AnCora website. Any license-related questions have to be directed to the original data providers at the University of Barcelona (that is, not to the UD contact address listed at the end of this README file).
The MISC column contains annotation of named entities and coreference, converted from the original XML files of AnCora-CO and merged with the UD-style morpho- syntactic annotation. The format of these annotations is described in Nedoluzhko et al. (2021): Coreference meets Universal Dependencies – a pilot experiment on harmonizing coreference datasets for 11 languages.
The following paper must be cited when using this corpus:
- Taulé, M., M.A. Martí, M. Recasens (2008) 'Ancora: Multilevel Annotated Corpora for Catalan and Spanish', Proceedings of 6th International Conference on Language Resources and Evaluation. Marrakesh (Morocco).
In addition, the following paper must be cited if coreference information (attributes entity, coreftype, corefsubtype, homophoricDD or entityref) is used:
- Recasens, Marta, M. Antònia Martí (2010) ‘AnCora-CO: Coreferentially annotated corpora for Spanish and Catalan’. Language Resources and Evaluation, Springer Science.
Additionally, the following paper must be cited when argumental attributes in "sn" or "grup.nom" (attributes func, arg, tem or lexicalid) are used:
- Peris, Aina, Mariona Taulé, Horacio Rodríguez (2010) ‘Semantic Annotation of Deverbal Nominalizations in the Spanish AnCora corpus’. Treebanks and Linguistic Theories (TLT-2010), Estonia.
- 2024-11-15 v2.15
- Removed feature AdvType.
- 2024-05-15 v2.14
- Dates ("el 7 d'abril") should be connected via nmod, not compound.
- Percentages ("40 per cent") should be connected via nmod, not compound.
- 2023-11-15 v2.13
- Combination of letters and digits is not NUM but SYM.
- NumForm used where appropriate.
- Currencies do not have NumForm=Digit (but they keep their special XPOS).
- Fixed: Sentence-initial "i" is cc, not advmod.
- Tokenization: number + "%" split into two tokens.
- 2023-05-15 v2.12
- Propagated the arg and tem attributes from the original AnCora.
- Per-nominals under participles changed from obj to obl:agent.
- Prepositional objects changed to obl:arg.
- Clitic doubling: only the noun is obj. The clitic changed to expl.
- Named entities that are not annotated for coreference marked as such.
- 2022-05-15 v2.10
- Some fixes in the LEMMA column.
- Heuristically resolved coreference cluster type mismatches.
- Coreference annotation converted to CorefUD 1.0 format.
- 2021-11-15 v2.9
- Changed sentence ids to reflect the original AnCora documents.
- Reordered sentences by sentence ids.
- Added document boundaries.
- Manually fixed numerous validation errors.
- Copied XPOS from the original AnCora.
- Copied CorefUD-style coreference and entity annotation from the original AnCora.
- Created enhanced representation (needed for coreference annotation).
- The license changed to CC BY 4.0 (https://doi.org/10.5281/zenodo.4762030).
- 2021-05-15 v2.8
- Manually fixed offset errors in lemmas inside MWEs (#2).
- Removed XPOS (which was uninformative).
- Cleaned up lemmas and features of personal pronouns.
- Split fused preposition-articles al(s), del(s), pel(s).
- 2019-05-15 v2.4
- Automatically fixed some validation errors.
- 2018-04-15 v2.2
- Repository renamed from UD_Catalan to UD_Catalan-AnCora.
- 2017-03-01 v2.0
- Converted to UD v2 guidelines.
- 2016-05-01 v1.3
- Initial release.
