Summary
The Lithuanian dependency treebank ALKSNIS v3.0 (Vytautas Magnus University).
Introduction
This is a conversion of the ALKSNIS treebank to Universal Dependencies. The original treebank was annotated in a style derived from the Prague Dependency Treebank of Czech. The original treebank is available at https://github.com/Semantika2/Alksnis-v.3.0. ALKSNIS v2.1 is published in the CLARIN LT repository at http://hdl.handle.net/20.500.11821/10. (Some users experience DNS errors when trying to access the repository; configuring the client machine to use 8.8.8.8 as the DNS server may help. See also http://clarin-lt.lt/?page_id=86.)
ALKSNIS v3 consists of 3,643 syntactically annotated sentences. Each node of a tree corresponds to a word, a punctuation mark or other text element (symbol, digit etc.) within a sentence. The following information is presented for each node: 1) a used form; 2) a lemma; 3) a morphology tag, and 4) a syntactic function (subject, object, etc.). Dependencies are shown by links between words.
The morphology tag set Jablonskis is used since ALKSNIS v2.2 and in the XPOS column of the UD conversion. This is a change from ALKSNIS 2 where a version of the MULTEXT-East tag set was used.
Syntactically annotated sentences are corrected according to guidelines that were created by scientists of VMU CCL, following rules of Prague Dependency Treebank. All the sentences are being manually checked and corrected by a group of linguists.
Acknowledgements
From v2.1 to v3.0 the treebank was developed within the project "Semantika2" (Nr. 02.3.1-CPVA-V-527-01-0002). The project was funded by European Structural Funds.
References
- Agnė Bielinskienė, Loïc Boizou, Jolanta Kovalevskaitė, Erika Rimkutė (2016): Lithuanian Dependency Treebank ALKSNIS. In: I. Skadiņa and R. Rozis (Eds.): Human Language Technologies – The Baltic Perspective, pp. 107–114. Amsterdam: IOS Press. doi:10.3233/978-1-61499-701-6-107 http://fcim.vdu.lt/~erika_rimkute/straipsniai/Alksnis_HLT.pdf, http://ebooks.iospress.nl/volumearticle/45523.
Changelog
- 2020-05-15 v2.6
- Genitive, dative and instrumental nominals are now considered oblique.
- Added enhanced relations with case information.
- Added enhanced relations around relative clauses.
- Added enhanced external subjects in control verb constructions.
- Added empty nodes to enhanced graphs (but orphans are just converted to dep).
- 2019-11-15 v2.5
- The existing data underwent full review of syntactic information based on improved guidelines to enhance annotation quality.
- Added new data: scientific abstracts and reviews, additional administrative texts.
- Jablonskis tagset is used instead of the MULTEXT-East tagset in the XPOS column.
- 2019-05-15 v2.4
- Initial UD release.
=== Machine-readable metadata (DO NOT REMOVE!) ================================ Data available since: UD v2.4 License: CC BY-SA 4.0 Includes text: yes Genre: news nonfiction legal fiction Lemmas: manual native UPOS: converted from manual XPOS: manual native Features: converted from manual Relations: converted from manual Contributors: Utka, Andrius; Rimkutė, Erika; Bielinskienė, Agnė; Kovalevskaitė, Jolanta; Boizou, Loïc; Aleksandravičiūtė, Gabrielė; Brokaitė, Kristina; Zeman, Daniel Contributing: elsewhere Contact: semantikalt@gmail.com, andrius.utka@vdu.lt, zeman@ufal.mff.cuni.cz ===============================================================================