UniversalDependencies / UD_Russian-Taiga

Other
12 stars 1 forks source link

Summary

Universal Dependencies treebank is based on data samples extracted from Taiga Corpus and MorphoRuEval-2017 and GramEval-2020 shared tasks collections.

Introduction

UD Russian Taiga has been developed at the School of Linguistics, National Research University Higher School of Economics in Moscow (HSE/Vyshka). The selection of texts is meant to represent those registers that have not been covered by UD Russian SynTagRus and UD Russian Google Stanford Dependencies, mainly e-communication (blogs and social media). The sentences are extracted from two open data collections. Taiga Corpus (https://tatianashavrina.github.io/taiga_site/) is an open-source corpus for machine learning collected by students as part of the curriculum of the MA Program in Computational Linguistics at HSE. MorphoRuEval 2017 text collections (https://github.com/dialogue-evaluation/morphoRuEval-2017) is an output of the RuEval shared task 'Evaluation of Russian NLP: Morphological analysis, http://www.dialog-21.ru/en/evaluation/2017/morphology/). GramEval 2020 collection (https://competitions.codalab.org/competitions/22902)[https://competitions.codalab.org/competitions/22902] is an output of the GramEval 2020 Shared Task on Russian Full Morphology and Dependency Parsing which consists of test data for five genres (social, wiki, news, fiction, poetry).

The plain text data were tokenized, lemmatized and parsed using UDpipe (http://ufal.mff.cuni.cz/udpipe) and checked manually. Corrections were made at all levels: tokenization, lemmata, pos, features, dependency relations.

Registers

Data split

Acknowledgments

We are grateful to all the contributors to the original open Russian data collections and especially to Tatiana Shavrina (Taiga, GramEval-2020) and Alena Fenogenova (MorphoRuEval-2017).

References

Changelog

=== Machine-readable metadata (DO NOT REMOVE!) ================================
Data available since: UD v2.2
License: CC BY-SA 4.0
Includes text: yes
Genre: blog fiction news poetry social wiki
Lemmas: manual native
UPOS: manual native
XPOS: manual native
Features: manual native
Relations: manual native
Contributors: Lyashevskaya, Olga; Rudina, Olga; Vlasova, Natalia; Zhuravleva, Anna
Contributing: elsewhere
Contact: olesar@yandex.ru
===============================================================================