Kompʹûternaâ 2021

10 papers
The 2021 edition of the Kompʹûternaâ conference featured nine papers primarily focused on advancing natural language processing (NLP) benchmarks and methodologies for the Russian language. A significant portion of the research centered on evaluation frameworks, including the RuSimpleSentEval-2021 shared task for sentence simplification and the RuShiftEval task for semantic shift detection. Additionally, the Russian SuperGLUE benchmark was updated to further assess the capabilities of Russian NLP models, with specific studies examining the effectiveness of rule-based heuristics compared to neural approaches. Technical contributions also explored the application of large-scale transformer models. Researchers investigated the use of Generative Pretrained Transformer-3 (GPT-3) for news clustering and title generation, as well as autoregressive models for text simplification. Other work addressed model optimization through data pseudo-labeling for multitask BERT adaptation and the evaluation of commonsense reasoning in conversational agents. Prolific contributors to this year's proceedings included Ekaterina Artemova, Valentin Malykh, and Mikhail Burtsev. These works reflect a trend toward standardizing evaluation metrics and adapting generative architectures for complex linguistic tasks in Russian.

Top Research Directions

The 2 research directions with Kompʹûternaâ 2021 papers, ranked by paper count, next to 2020. Share is the fraction of the edition’s papers in the direction.

#Direction20212020Share 2021Share 20202020 rank
1Shifting from Static Benchmarks to Dynamic Evaluation1010.0%0.0%—
2The Multilingual Evaluation Gap1010.0%0.0%—

Download all directions as CSV

Institutions

Institutions credited by Kompʹûternaâ 2021 papers: a paper credits every institution named in its affiliations, once. 6 institutions share 4 credited papers.

#InstitutionPapersShare
1Moscow Institute of Physics and Technology 375.0%
2National Research University Higher School of Economics 375.0%
3AIRI - Artificial Intelligence Research Institute 250.0%
4Huawei Technologies (China) 250.0%
5Kazan Federal University 250.0%
6University of Oslo 125.0%

Download all institutions as CSV

About this data

Papers. Kompʹûternaâ 2021 lists 10 papers on Lacuna, duplicate versions counted once; rejected and withdrawn submissions are left out.

Directions. Directions are Lacuna’s clustering of the concept elements extracted from each paper, so a paper can belong to several and direction counts add up to more than the paper count. A direction’s share is its papers divided by the edition’s papers. Comparisons with 2020 use shares, since the edition went from 6 to 10 papers.

Institutions. A paper credits each institution named in its affiliations, resolved to a ROR record, once. 4 of 10 papers (40.0%) credit at least one institution and are the base for institution shares; the rest have no resolved affiliation yet.

Data. Counts come from Lacuna’s data build of 2026-11-22 and change as papers are added. Tables as CSV: directions · institutions. Cite as: Lacuna, “Kompʹûternaâ 2021”, https://lacuna.tiptreesystems.com/venue/komputernaa/2021.

Top Authors

AuthorPapers
Mikhail Burtsev2
Elena Tutubalina1

Papers

Title
Data pseudo-labeling while adapting BERT for multitask approaches
Evaluation of Conversational Skills for Commonsense
RuShiftEval: a shared task on semantic shift detection for Russian
RuSimpleSentEval-2021 Shared Task: Evaluating Sentence Simplification for Russian
Russian SuperGLUE 1.1: Revising the Lessons not Learned by Russian NLP models
Russian SuperGLUE 1.1: Revising the Lessons not Learned by Russian NLP-models
Short Text Clustering with Transformers
Text Simplification with Autoregressive Models
Unreasonable Effectiveness of Rule-Based Heuristics in Solving Russian SuperGLUE Tasks
Using Generative Pretrained Transformer-3 Models for Russian News Clustering and Title Generation tasks