Grameval 2020 Shared Task: Russian Full Morphology and Universal Dependencies Parsing

Lyashevskaya, O. N.; Shavrina, Tatiana; Трофимов, Игорь Владимирович; Vlasova, Natalia; Sberbank, Moscow

doi:10.28995/2075-7182-2020-19-553-569

Cited by 15 publications

(8 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…We compare LIMA performance on Russian-SynTagRus corpus using the official CoNLL 2018 evaluation script and on GramEval-2020 [14] corpus using its official evaluation script. The use of the evaluation script from CoNLL 2018 competition is motivated by the intention to compare our results with previous works.…”

Section: Discussionmentioning

confidence: 99%

The Russian Language Pipeline in the Lima Multilingual Analyzer

Bocharov¹,

Chalendar²

2020

Computational Linguistics and Intellectual Technologies

View full text Add to dashboard Cite

In this paper we describe the implementation of Russian language pipeline in LIMA multilingual analyzer and the results obtained in GramEval-2020 shared task. LIMA is a modular pipeline that implements rule-based and machine learning analysis components. Russian language pipeline includes deep neural networks based modules for tokenization, sentence segmentation, part of speech tagging, lemmatization and dependency parsing. Part of speech tags, feature tags and dependency trees conform to Universal Dependencies rules.

show abstract

Section: Discussionmentioning

confidence: 99%

The Russian Language Pipeline in the Lima Multilingual Analyzer

Bocharov¹,

Chalendar²

2020

Computational Linguistics and Intellectual Technologies

View full text Add to dashboard Cite

show abstract

“…At the MorphoRuEval-2017 shared task [11], a 96.91% accuracy score in lemmatization was achieved on a balanced set of data from various sources (news, social networks, fiction, etc.). And in the GramEval-2020 shared task [12] the track became even more complicated since data from social media, poetry and historical texts of the 17th century were added to the test sample: the best overall lemmatization score being 98% on fiction texts, 98.2% on the news, 95.3% on poetry, 96% on social media, 93% on wiki and 78.3% on historical texts. It became manifest that it is technically possible for the Russian language to pose more complex challenges, especially for notoriously "difficult-to-process" groups of words and lexical categories.…”

Section: Previous Workmentioning

confidence: 99%

Using RuGPT3-XL Model for RuNormAS competition

Anton¹,

Shliazhko²,

Katricheva³

et al. 2021

Computational Linguistics and Intellectual Technologies

View full text Add to dashboard Cite

The paper presents a fine-tuning methodology of the RuGPT3-XL (Generative Pretrained Transformer-3 for Russian) language model for the normalization of text spans task. The solution is presented in a competition for two tasks: Normalization of Named Entities (Named entities) and Normalization of a wider class of text spans, including the normalization of different parts of speech (Generic spans). The best solution has achieved 0.9645 accuracy on the Generic spans task and 0.9575 on the Named entities task. The presented solutions are in the public domain at https://github.com/RussianNLP/RuNormAS-solution

show abstract

“…The model we used in our approach closely follows the model from [1] which has shown the stateof theart performance in morphosyntactic parsing on the GramEval2020 dataset [6].…”

Section: Modelmentioning

confidence: 99%

“…https://natasha.github.io/ 4 "The Government of the Russian Federation will consider…" 5 "Russian"6 "The Russian Federation"…”

mentioning

confidence: 99%

Annotated Span Normalization as a Sequence Labelling Task

Anastasyev¹,

Yandex²

2021

Computational Linguistics and Intellectual Technologies

View full text Add to dashboard Cite

In this paper, we describe a way to perform span normalization as a sequence labelling task. Our model predicts the modifications that should be applied to the span tokens to normalize them. This prediction is performed via se quence labelling, which means that each token is normalized independently. Despite the simplicity of the approach, we show that it can lead to the stateoftheart results. We compare different pretraining schemas in application to this task. We show that the best quality can be achieved when the normalizer is trained on top of a BERTbased morphosyntactic parser's representations. Moreover, we propose some additional features useful in the task and prove that auxiliary morphosyntactic losses can help the model. Furthermore, we show that the model compares favourably with other contestant models of the RuNormAS competition.

show abstract

Grameval 2020 Shared Task: Russian Full Morphology and Universal Dependencies Parsing

Cited by 15 publications

References 3 publications

The Russian Language Pipeline in the Lima Multilingual Analyzer

The Russian Language Pipeline in the Lima Multilingual Analyzer

Using RuGPT3-XL Model for RuNormAS competition

Annotated Span Normalization as a Sequence Labelling Task

Contact Info

Product

Resources

About