Summary

Czech-PDT is based on the written portion of Prague Dependency Treebank - Consolidated 1.0 (PDT-C), created at the Charles University, Prague.

References

@inproceedings{pdt-c,
  author    = {Jan Hajič and Eduard Bejček and Jaroslava Hlaváčová and Marie Mikulová and Milan Straka and Jan Štěpánek and Barbora Štěpánková},
  title     = {Prague Dependency Treebank - Consolidated 1.0},
  booktitle = {Proceedings of the 12th International Conference on Language Resources and Evaluation (LREC 2020)},
  publisher = {European Language Resources Association},
  address   = {Marseille, France},
  year      = 2020,
  pages     = {5208--5218},
}

Changelog

2024-03-28 v1.2

Improved distinction between adverbial predicates (with copula) and adverbial modifiers.
Coreference annotation: If a bracket is in mention span, the paired bracket is added too, if possible.
More restrictive use of orphans and empty nodes: Not in non-verbal coordinated sentences.
Fixed crossing coreference mentions.
Fixed treatment of "by" in aux/cop chains.
Improved form and position of abstract predicates in gapping.
Removed NumValue from all Czech UD treebanks.
Pseudo-existential "být" with oblique/adverbial modifiers changed to copula.
Source data switched from PDT 3.0 to PDT-C 1.0.
- Underlying text data is the same.
- Changed some aspects of lemmatization, including LId and other attributes in MISC.
- Somewhat different XPOS tag set.
- UD features: now all verbs have Aspect; minor changes at various other places.
- Foreign words are now systematically tagged X (previously, many of them had descriptive UPOS tags).
The tectogrammatical (t-) layer of source annotation is now used for documents for which it is available.
- Sentences converted with the help of t-layer have the comment "Tectogrammatical annotation available."
- There are more enhanced dependency relations and empty nodes.
- The MISC column contains tectogrammatical functors.
Temporary fix of double subjects (second subject converted to dep). In the long run, the cause should be found and fixed upstream.
Added the enhanced relation subtype nsubj:xsubj.

2023-02-24 v1.1

Added VerbForm=Part|Voice=Pass to long forms of passive participles.
Added VerbForm=Vnoun to verbal nouns.
The verb 'být' is now AUX in all contexts.
Merged PRON/DET 'sám', 'samý'.
Removed superfluous empty nodes #Rcp, #Cor, #QCor.
Removed empty nodes depending on the artificial 0:root.
"Bych/bys/by/bychom/byste" in MWTs no longer breaks mention spans.
Improved guessing of pronominal forms for empty nodes.
Functors added also to non-empty nodes.
Temporary fix of double subjects (second subject converted to dep). In the long run, the cause should be found and fixed upstream.
Added the enhanced relation subtype nsubj:xsubj.

2022-04-06 v1.0

new format of coreference and anaphora annotations
Improved conversion from the Prague format, reduced number of discontinuous mentions.
Case-enhanced dependency relations made more consistent.

2021-12-10 v0.2

Fixed attachment of punctuation where required by the UD guidelines.

2021-03-11 v0.1

initial conversion

=== Machine-readable metadata (DO NOT REMOVE!) ================================
Data available since: CorefUD 0.1
License: CC BY-NC-SA 4.0
Includes text: yes
Genre: news reviews nonfiction
Lemmas: converted from manual
UPOS: converted from manual
XPOS: manual native
Features: converted from manual
Relations: converted from manual
CorefUD contributors: Zeman, Daniel (1); Hajič, Jan (1); Straka, Milan (1)
Other contributors: Bejček, Eduard; Hlaváčová, Jaroslava; Mikulová, Marie; Štěpánek, Jan; Štěpánková, Barbora
Contributors' affiliations: (1) Charles University, Faculty of Mathematics and Physics, Institute of Formal and Applied Linguistics, Prague, Czechia
Contributing: elsewhere
Contact: [email protected]
===============================================================================