Papers by Alexander Panchenko

Studying Taxonomy Enrichment on Diachronic WordNet Versions (2020.coling-main)

Challenge:	Existing studies assume that all facts are equally forgettable . popular facts, frequent and widely distributed, may be more deeply embedded than rare ones, making them harder to erase.
Approach:	They propose a benchmark to evaluate how unlearning differs between pretrained and supervised fine-tuned models when fact popularity is taken into account.
Outcome:	The proposed model is compared with pretrained and SFT models on the forget data and shows that it performs better on both models.

Copied to clipboard

Irina Nikishina, Varvara Logacheva, Alexander Panchenko, Natalia Loukachevitch

Which is Better for Deep Learning: Python or MATLAB? Answering Comparative Questions in Natural Language (2021.eacl-demos)

Challenge:	Ontologies, taxonomies and thesauri are used in many NLP tasks but are often not maintained.
Approach:	They propose methods for taxonomy enrichment in a resource-poor setting . they also create novel datasets for training and evaluating taxonomies .
Outcome:	The proposed methods are applicable to English and Russian datasets and can be used in other languages.

Copied to clipboard

Viktoriia Chekalina, Alexander Bondarenko, Chris Biemann, Meriem Beloucif, Varvara Logacheva, Alexander Panchenko

Natural Language Interfaces

Challenge:	Comparative QA is a challenging task since it requires collecting evidence from many different sources.
Approach:	They propose a natural language interface for comparative QA that can be used in personal assistants, chatbots, and similar NLP devices.
Outcome:	The proposed system can be used in personal assistants, chatbots, and similar NLP devices.

Boosting Self-Consistency with Ranking (2026.acl-srw)

Copied to clipboard

Maria Marina, Daniil Moskovskiy, Sergey Pletenev, Mikhail Salnikov, Alexander Panchenko, Viktor Moskvoretskii

MultiParaDetox: Extending Text Detoxification with Parallel Data to New Languages (2024.naacl-short)

Challenge:	Existing approaches to improve performance of large language models include self-consistency, RISC, extended reasoning, and iterative self-correction.
Approach:	They propose a test-time scaling technique that uses multiple features to score candidate answers in self-consistency as a ranking problem.
Outcome:	The proposed method achieves better accuracy-efficiency trade-off than standard self-consistency and strong baselines on three datasets.

Copied to clipboard

Daryna Dementieva, Nikolay Babakov, Alexander Panchenko

Challenge:	Text detoxification is a textual style transfer task where a toxic text is paraphrased to the neutral register.
Approach:	They propose to extend ParaDetox pipeline to multiple languages to automate parallel detoxification corpus collection.
Outcome:	The proposed methods have been used in toxic speech combating and toxic speech fighting tasks.

Adaptive Retrieval Without Self-Knowledge? Bringing Uncertainty Back Home (2025.acl-long)

Copied to clipboard

Viktor Moskvoretskii, Maria Marina, Mikhail Salnikov, Nikolay Ivanov, Sergey Pletenev, Daria Galimzianova, Nikita Krayko, Vasily Konovalov, Irina Nikishina, Alexander Panchenko

Question Answering

Challenge:	Recent adaptive retrieval methods integrate LLMs’ intrinsic knowledge with external information appealing to LLM self-knowledge, but they often neglect efficiency evaluations and comparisons with uncertainty estimation techniques.
Approach:	They propose to integrate LLMs’ intrinsic knowledge with external information appealing to LLM self-knowledge but neglect efficiency evaluations and comparisons with uncertainty estimation techniques.
Outcome:	The proposed methods outperform complex pipelines in terms of efficiency and self-knowledge while maintaining comparable QA performance.

Making Fast Graph-based Algorithms with Graph Metric Embeddings (P19-1)

Copied to clipboard

Andrey Kutuzov, Mohammad Dorgham, Oleksiy Oliynyk, Chris Biemann, Alexander Panchenko

Token-Level Density-Based Uncertainty Quantification Methods for Eliciting Truthfulness of Large Language Models (2025.naacl-long)

Challenge:	Graph measures, such as node distances, are inefficient to compute.
Approach:	They propose a way to learn graph embeddings by using vector operations instead of a graph structure.
Outcome:	The proposed method outperforms other graph embeddings on word similarity and word sense disambiguation tasks.

Copied to clipboard

Artem Vazhentsev, Lyudmila Rvanova, Ivan Lazichny, Alexander Panchenko, Maxim Panov, Timothy Baldwin, Artem Shelmanov

How Certain is Your Transformer? (2021.eacl-main)

Challenge:	Uncertainty quantification (UQ) is a prominent approach for eliciting truthful answers from large language models (LLMs).
Approach:	They propose to use a well-established method for text generation to extract token embeddings from multiple layers of LLMs and compute MD scores for each token.
Outcome:	The proposed method improves on existing methods and provides accurate and computationally efficient uncertainty scores for sequence-level selective generation and claim-level fact-checking tasks.

Copied to clipboard

Artem Shelmanov, Evgenii Tsymbalov, Dmitri Puzyrev, Kirill Fedyanin, Alexander Panchenko, Maxim Panov

MEKER: Memory Efficient Knowledge Embedding Representation for Link Prediction and Question Answering (2022.acl-srw)

Challenge:	Obtaining reliable uncertainty estimations for such neural networks (NNs) is challenging due to the huge number of parameters in these deep learning models.
Approach:	They propose to use Monte Carlo dropout to estimate uncertainty for Transformer-based models and construct inexpensive estimates using Determinantal Point Processes.
Outcome:	The proposed estimates improve the quality of detection of error-prone instances.

Copied to clipboard

Viktoriia Chekalina, Anton Razzhigaev, Albert Sayapin, Evgeny Frolov, Alexander Panchenko

Documents Representation via Generalized Coupled Tensor Chain with the Rotation Group constraint (2021.findings-acl)

Challenge:	Existing methods to embed learning use a standard Neural Networks (NN) backward mechanism, duplicating its memory consumption.
Approach:	They propose a memory-efficient KG embedding model that embeds knowledge graphs as 3rd-order binary tensors.
Outcome:	The proposed model yields comparable performance on link prediction and KG-based question answering tasks.

Copied to clipboard

Igor Vorona, Anh-Huy Phan, Alexander Panchenko, Andrzej Cichocki

Will It Still Be True Tomorrow? Multilingual Evergreen Question Classification to Improve Trustworthy QA (2025.emnlp-main)

Challenge:	despite the diversity of linguistic structures, vector embedding models lack order-preserving properties . current methods for learning linguistic structure can be expensive and time-consuming .
Approach:	They propose a method for embedding documents and words in rotation group . they capture word order and higher-order word interactions .
Outcome:	The proposed model achieves the best results in document classification benchmarks.

Copied to clipboard

Sergey Pletenev, Maria Marina, Nikolay Ivanov, Daria Galimzianova, Nikita Krayko, Mikhail Salnikov, Vasily Konovalov, Alexander Panchenko, Viktor Moskvoretskii

Question Answering

Challenge:	Large Language Models (LLMs) often struggle with question answering due to hallucinated answers.
Approach:	They propose a multilingual QA dataset with evergreen labels that can be used to evaluate and train large language models.
Outcome:	The proposed model performs well on 12 modern LLMs and EG-E5 classifiers.

Hybrid Uncertainty Quantification for Selective Text Classification in Ambiguous Tasks (2023.acl-long)

Copied to clipboard

Artem Vazhentsev, Gleb Kuzmin, Akim Tsvigun, Alexander Panchenko, Maxim Panov, Mikhail Burtsev, Artem Shelmanov

LM-Polygraph: Uncertainty Estimation for Language Models (2023.emnlp-demo)

Challenge:	Existing methods for text classification tasks are inherently ambiguous and can cause errors.
Approach:	They propose a method that combines epistemic and aleatoric uncertainty to estimate toxicity detection errors.
Outcome:	The proposed method outperforms existing methods for toxicity detection and other ambiguous text classification tasks.

Copied to clipboard

Ekaterina Fadeeva, Roman Vashurin, Akim Tsvigun, Artem Vazhentsev, Sergey Petrakov, Kirill Fedyanin, Daniil Vasilev, Elizaveta Goncharova, Alexander Panchenko, Maxim Panov, Timothy Baldwin, Artem Shelmanov

Uncertainty Estimation of Transformer Predictions for Misclassification Detection (2022.acl-long)

Challenge:	Large language models often "hallucinate" i.e., fabricate facts without providing users an apparent means to discern the veracity of their statements.
Approach:	They propose a framework with implementations of state-of-the-art UE methods for LLMs with unified program interfaces in Python.
Outcome:	The proposed framework implements state-of-the-art UE methods for LLMs with unified program interfaces in Python and an extendable benchmark for consistent evaluation by researchers.

Copied to clipboard

Artem Vazhentsev, Gleb Kuzmin, Artem Shelmanov, Akim Tsvigun, Evgenii Tsymbalov, Kirill Fedyanin, Maxim Panov, Alexander Panchenko, Gleb Gusev, Mikhail Burtsev, Manvel Avetisian, Leonid Zhukov

Efficient Out-of-Domain Detection for Sequence to Sequence Models (2023.findings-acl)

Challenge:	Uncertainty estimation (UE) of model predictions is crucial step for a variety of tasks such as active learning, misclassification detection, adversarial attack detection, etc.
Approach:	They propose to modify UE methods for Transformer models for misclassification detection in named entity recognition and text classification tasks to improve model expressiveness and computational performance.
Outcome:	The proposed methods outperform computationally intensive methods on misclassification detection tasks and are based on a large dataset of simulated datasets.

Copied to clipboard

Artem Vazhentsev, Akim Tsvigun, Roman Vashurin, Sergey Petrakov, Daniil Vasilev, Maxim Panov, Alexander Panchenko, Artem Shelmanov

Text Generation

Challenge:	Sequence-to-sequence (seq2sequ) models are a ubiquitous tool for text generation but they are not suitable for many other tasks.
Approach:	They propose to use UE techniques to identify out-of-domain (OOD) inputs where the model is susceptible to errors.
Outcome:	The proposed methods outperform heavyweight ensembles on the task of OOD detection.

Are Large Language Models Good at Lexical Semantics? A Case of Taxonomy Learning (2024.lrec-main)

Copied to clipboard

Viktor Moskvoretskii, Alexander Panchenko, Irina Nikishina

Unconditional Truthfulness: Learning Unconditional Uncertainty of Large Language Models (2025.emnlp-main)

Challenge:	Recent studies on LLMs do not pay enough attention to linguistic and lexical semantic tasks, such as taxonomy learning.
Approach:	They propose a method for stochastic graph traversal and a new algorithm for data collection . they propose LLaMA-2 and Mistral for a lexical semantic task .
Outcome:	The proposed models can perform linguistic and lexical tasks, but they lack basic skills in taxonomy learning.

Copied to clipboard

Artem Vazhentsev, Ekaterina Fadeeva, Rui Xing, Gleb Kuzmin, Ivan Lazichny, Alexander Panchenko, Preslav Nakov, Timothy Baldwin, Maxim Panov, Artem Shelmanov

HatePRISM: Policies, Platforms, and Research Integration. Advancing NLP for Hate Speech Proactive Mitigation (2025.findings-acl)

Challenge:	Uncertainty quantification (UQ) is a promising approach for detecting hallucinations and low-quality outputs of Large Language Models (LLMs).
Approach:	They propose to learn conditional dependency between autoregressive LLM generation steps from attention-based features and a two-staged training procedure to incorporate recurrent features.
Outcome:	The proposed method is highly effective for selective generation, achieving substantial improvements over rivaling unsupervised and supervised approaches.

Copied to clipboard

Naquee Rizwan, Seid Muhie Yimam, Daryna Dementieva, Dr. Florian Skupin, Tim Fischer, Daniil Moskovskiy, Aarushi Ajay Borkar, Robert Geislinger, Punyajoy Saha, Sarthak Roy, Martin Semmann, Alexander Panchenko, Chris Biemann, Animesh Mukherjee

On the Compositionality Prediction of Noun Phrases using Poincaré Embeddings (P19-1)

Challenge:	Existing approaches to manage hate speech rely on reactive measures such as blocking or suspending offensive messages . despite regulations imposed by nations and social media platforms, hateful content remains a challenge .
Approach:	They propose a framework for automated hate speech moderation based on different strategies . they examine hate speech regulations and strategies from three perspectives .
Outcome:	The proposed framework could be based on a combination of country regulations, social platform policies, and NLP research datasets.

Copied to clipboard

Abhik Jana, Dima Puzyrev, Alexander Panchenko, Pawan Goyal, Chris Biemann, Animesh Mukherjee

LLM-Independent Adaptive RAG: Let the Question Speak for Itself (2025.emnlp-main)

Challenge:	idiomatic phrases have a non-compositional meaning, meanings of which can be derived from constituents and their grammatical relations.
Approach:	They propose to combine hierarchical and distributional information to blend hierarchic and distribution-based hierarchies to detect compositionality for noun phrases.
Outcome:	The proposed technique achieves significant improvements over state-of-the-art models based on distributional information and a weighted average of the distributional similarity and p-like function.

Copied to clipboard

Maria Marina, Nikolay Ivanov, Sergey Pletenev, Mikhail Salnikov, Daria Galimzianova, Nikita Krayko, Vasily Konovalov, Alexander Panchenko, Viktor Moskvoretskii

SkipCLM: Enhancing Crosslingual Alignment of Decoder Transformer Models via Contrastive Learning and Skip Connection (2025.naacl-srw)

Challenge:	Existing methods to retrieve Large Language Models (LLMs) are inefficient and impractical.
Approach:	They propose a lightweight adaptive retrieval method that leverages external information to achieve comparable quality while achieving significant efficiency gains.
Outcome:	The proposed methods achieve comparable quality while achieving significant efficiency gains on 6 QA datasets.

Copied to clipboard

Nikita Sushko, Alexander Panchenko, Elena Tutubalina

LLMs to Replace Crowdsourcing For Parallel Data Creation? The Case of Text Detoxification (2024.findings-emnlp)

Challenge:	Existing contrastive learning methods for cross-lingual alignment are not effective for multilingual machine translation tasks.
Approach:	They propose a method that augments contrastive learning for cross-lingual alignment with a trainable skip connection to preserve information crucial for accurate target language generation.
Outcome:	Experiments with XGLM-564M on the Flores-101 benchmark show that the proposed method preserves crucial information crucial for accurate target language generation.

Copied to clipboard

Daniil Moskovskiy, Sergey Pletenev, Alexander Panchenko

DimABSA: Building Multilingual and Multidomain Datasets for Dimensional Aspect-Based Sentiment Analysis (2026.acl-long)

Challenge:	Manual annotation methods, such as crowdsourcing, are costly and require intricate task design skills.
Approach:	They propose to use open source LLMs to annotate parallel data for text detoxification . they generate a pseudo-parallel detoxification dataset using activation patching .
Outcome:	The proposed model performs comparable to the original dataset in automatic detoxification evaluation metrics and superior quality in manual evaluation and side-by-side comparisons.

Copied to clipboard

Lung-Hao Lee, Liang-Chih Yu, Natalia V Loukachevitch, Ilseyar Alimova, Alexander Panchenko, Tzu-Mi Lin, Zhe-Yu Xu, Jian-Yu Zhou, Guangmin Zheng, Jin Wang, Sharanya Awasthi, Jonas Becker, Jan Philip Wahle, Terry Ruas, Shamsuddeen Hassan Muhammad, Saif M. Mohammad

Sentiment Analysis

Challenge:	Existing ABSA research relies on coarse-grained categorical labels, which limits its ability to capture nuanced affective states.
Approach:	They propose a dimensional approach that represents sentiment with continuous valence–arousal (VA) scores, enabling fine-grained analysis at both the aspect and sentiment levels.
Outcome:	The proposed approach represents sentiment with continuous valence–arousal (VA) scores, enabling fine-grained analysis at both the aspect and sentiment levels.

The Silence of the Facts: Popularity as a Barrier to Machine Unlearning (2026.acl-srw)

Copied to clipboard

Anna Borisiuk, Andrey Savchenko, Alexander Panchenko, Elena Tutubalina

Challenge:	Existing unlearning methods assume that all facts are equally challenging to forget . large models struggle more to forget popular entities, damaging related knowledge in the process .
Approach:	They build a benchmark to investigate whether fact popularity influences the efficiency of LLM unlearning.
Outcome:	The proposed benchmark compares state-of-the-art models on a set of models of different sizes.

Improving Hypernymy Extraction with Distributional Semantic Classes (L18-1)

Copied to clipboard

Alexander Panchenko, Dmitry Ustalov, Stefano Faralli, Simone P. Ponzetto, Chris Biemann

TaxFree: a Visualization Tool for Candidate-free Taxonomy Enrichment (2022.aacl-demo)

Challenge:	Existing methods for extracting hypernyms focus on the acquisition of binary hypernies .
Approach:	They propose a distributionally-induced semantic class for extracting hypernyms . they also use distributional semantics to induce sense-aware semantic classes .
Outcome:	The proposed method improves the quality of the hypernymy extraction in terms of precision and recall.

Copied to clipboard

Irina Nikishina, Ivan Andrianov, Alsu Vakhitova, Alexander Panchenko

Challenge:	In this paper, we present an open source system for taxonomy visualisation and automatic taxonomies enrichment without pre-defined candidates.
Approach:	They propose an open source system for taxonomy visualisation and automatic taxonomie enrichment without pre-defined candidates on the example of WordNet-3.0.
Outcome:	The proposed system can be used for visualisation and inspection of taxonomies without pre-defined candidates on WordNet-3.0.

SynthDetoxM: Modern LLMs are Few-Shot Parallel Detoxification Data Annotators (2025.naacl-long)

Copied to clipboard

Daniil Moskovskiy, Nikita Sushko, Sergey Pletenev, Elena Tutubalina, Alexander Panchenko

Multilinguality

Challenge:	Existing approaches to multilingual text detoxification are hampered by the scarcity of parallel multilingual datasets.
Approach:	They propose a pipeline for the generation of multilingual parallel detoxification data and a dataset for SynthDetoxM which is manually generated and rewritten with open-source LLMs.
Outcome:	The proposed pipeline outperforms human-annotated datasets even in data limited setting.

Always Keep your Target in Mind: Studying Semantics and Improving Performance of Neural Lexical Substitution (2020.coling-main)

Copied to clipboard

Nikolay Arefyev, Boris Sheludko, Alexander Podolskiy, Alexander Panchenko

Active Learning for Abstractive Text Summarization (2022.findings-emnlp)

Challenge:	Lexical substitution is a powerful technology used in various NLP applications . it generates plausible words that can replace a given word in a textual context .
Approach:	They propose to use a large-scale comparative study to compare lexical substitution methods . they compare existing and new methods using word sense induction datasets .
Outcome:	The proposed methods improve competitive results by incorporating information about the target word into the models.

Copied to clipboard

Akim Tsvigun, Ivan Lysenko, Danila Sedashov, Ivan Lazichny, Eldar Damirov, Vladimir Karlov, Artemy Belousov, Leonid Sanochkin, Maxim Panov, Alexander Panchenko, Mikhail Burtsev, Artem Shelmanov

RuPAWS: A Russian Adversarial Dataset for Paraphrase Identification (2022.lrec-1)

Challenge:	Abstractive text summarization (ATS) requires a long document and short summaries.
Approach:	They propose a query strategy for AL in abstractive text summarization that uses uncertainty estimation to reduce model performance.
Outcome:	The proposed query strategy improves ROUGE and consistency scores for annotated datasets . it also increases the performance of the model, compared to passive annotation.

Copied to clipboard

Nikita Martynov, Irina Krotova, Varvara Logacheva, Alexander Panchenko, Olga Kozlova, Nikita Semenov

Robustness in NLP

Challenge:	Existing datasets for paraphrase identification lack challenging sentence pairs with high word overlap.
Approach:	They propose to use a dataset for Russian paraphrase detection that includes examples from PAWS translated to the Russian language and manually annotated by native speakers.
Outcome:	The proposed model performs well on both datasets while maintaining accuracy on the ParaPhraser benchmark.

Fact-Checking the Output of Large Language Models via Token-Level Uncertainty Quantification (2024.findings-acl)

Copied to clipboard

Ekaterina Fadeeva, Aleksandr Rubashevskii, Artem Shelmanov, Sergey Petrakov, Haonan Li, Hamdy Mubarak, Evgenii Tsymbalov, Gleb Kuzmin, Alexander Panchenko, Timothy Baldwin, Preslav Nakov, Maxim Panov

TARGER: Neural Argument Mining at Your Fingertips (P19-3)

Challenge:	Large language models are notorious for producing erroneous claims in their output.
Approach:	They propose a fact-checking and hallucination detection pipeline based on token-level uncertainty quantification that removes the impact of uncertainty about what claim to generate on the current step and what surface form to use.
Outcome:	The proposed method can fact-check the atomic claims in the output of large language models.

Copied to clipboard

Artem Chernodub, Oleksiy Oliynyk, Philipp Heidenreich, Alexander Bondarenko, Matthias Hagen, Chris Biemann, Alexander Panchenko

TaxoLLaMA: WordNet-based Model for Solving Multiple Lexical Semantic Tasks (2024.acl-long)

Challenge:	Argumentation is a multi-disciplinary field that extends from philosophy and psychology to linguistics as well as to artificial intelligence.
Approach:	They propose to use TARGER to tagging arguments in free text and keyword-based retrieval of arguments from a web-scale corpus.
Outcome:	The proposed framework can be used without any reproducibility effort on the user's side and is easily portable to other domains and use cases.

Copied to clipboard

Viktor Moskvoretskii, Ekaterina Neminova, Alina Lobanova, Alexander Panchenko, Irina Nikishina

Text Detoxification using Large Pre-trained Neural Models (2021.emnlp-main)

Challenge:	Recent studies in Natural Language Processing widely utilize Large Language Models (LLMs) for their capability to store extensive knowledge.
Approach:	They propose an LLM-based model that captures lexical-semantic knowledge from WordNet and test it on multiple lexicals.
Outcome:	The proposed model achieves 11 SOTA results and 4 top-2 results out of 16 taxonomy-related tasks.

Copied to clipboard

David Dale, Anton Voronov, Daryna Dementieva, Varvara Logacheva, Olga Kozlova, Nikita Semenov, Alexander Panchenko

Evolutionary Search for Automated Design of Uncertainty Quantification Methods (2026.acl-srw)

Challenge:	Existing studies on text detoxification cast this task as style transfer . text detox requires better preservation of the original meaning, authors argue .
Approach:	They propose two unsupervised methods for eliminating toxicity in text . they use a paraphraser guided by style-trained language models to keep the text content .
Outcome:	The proposed methods yield new SOTA results.

Copied to clipboard

Mikhail Seleznyov, Daniil Korbut, Viktor Moskvoretskii, Oleg Somov, Alexander Panchenko, Elena Tutubalina

Explainability & Interpretability in NLP

Challenge:	Qualitative analysis reveals that different LLMs employ qualitatively distinct evolutionary strategies for automating, interpretable hallucination detector design.
Approach:	They apply LLM-powered evolutionary search to discover unsupervised UQ methods represented as Python programs and apply them to atomic claim verification.
Outcome:	The proposed methods outperform strong manually-designed baselines while generalizing robustly out-of-distribution.

Enriching Frame Representations with Distributionally Induced Senses (L18-1)

Copied to clipboard

Stefano Faralli, Alexander Panchenko, Chris Biemann, Simone Paolo Ponzetto

Detecting Overflow in Compressed Token Representations for Retrieval-Augmented Generation (2026.eacl-srw)

Challenge:	lexical resource that enriches Framester knowledge graph with semantic features from text corpora . paves way for development of novel, deeper semantic-aware applications .
Approach:	They propose a lexical resource that enriches the Framester knowledge graph with semantic features from text corpora.
Outcome:	The proposed resource enables the development of deeper semantic-aware applications . it combines knowledge from text and symbolic representations of events and participants .

Copied to clipboard

Julia Belikova, Danila Rozhevskii, Dennis Svirin, Konstantin Polev, Alexander Panchenko

Kandinsky: An Improved Text-to-Image Synthesis with Image Prior and Latent Diffusion (2023.emnlp-demo)

Challenge:	Efficient long-context processing remains a challenge for large language models (LLMs) however, the limits of compressibility remain underexplored.
Approach:	They propose a method to characterize and detect token overflow in xRAG soft-compression by mapping long contexts into dense vectors that can be directly consumed by the model.
Outcome:	The proposed method identifies token overflow with query-agnostic saturation statistics but lacks the capability to detect it.

Copied to clipboard

Anton Razzhigaev, Arseniy Shakhmatov, Anastasia Maltseva, Vladimir Arkhipkin, Igor Pavlov, Ilya Ryabov, Angelina Kuts, Alexander Panchenko, Andrey Kuznetsov, Denis Dimitrov

Multimodality

Challenge:	Experimental evaluations demonstrate FID score of 8.03 on the COCO-30K dataset, marking our model as the top open source performer in terms of measurable image generation quality.
Approach:	They propose a latent diffusion-based model that combines image prior and latent diffusive techniques to create a text-to-image architecture.
Outcome:	The proposed model achieves the highest FID score among open-source models . it is compared with the state-of-the-art models on the COCO-30K dataset .

Active Learning for Sequence Tagging with Deep Pre-trained Models and Bayesian Uncertainty Estimates (2021.eacl-main)

Copied to clipboard

Artem Shelmanov, Dmitri Puzyrev, Lyubov Kupriyanova, Denis Belyakov, Daniil Larionov, Nikita Khromov, Olga Kozlova, Ekaterina Artemova, Dmitry V. Dylov, Alexander Panchenko

Exploring Cross-lingual Text Detoxification with Large Multilingual Language Models. (2022.acl-srw)

Challenge:	Annotating training data for sequence tagging of texts is usually very time-consuming . active learning can help to reduce the amount of annotation required to train a good model by multiple times .
Approach:	They are the first to thoroughly investigate active learning and transfer learning for natural language processing . they propose to combine active learning with active learning to improve model acquisition .
Outcome:	The proposed combination of active learning and Bayesian uncertainty estimation improves performance and reduces obstacles for applying it in practice.

Copied to clipboard

Daniil Moskovskiy, Daryna Dementieva, Alexander Panchenko

Multilingual and Explainable Text Detoxification with Parallel Corpora (2025.coling-main)

Challenge:	Existing methods of textual style transfer are monolingual i.e. designed to work in one exact language.
Approach:	They propose to make large multilingual models capable of performing multilingual style transfer without direct fine-tuning in a given language.
Outcome:	The proposed model can generate text in a given language without fine-tuning and is able to perform cross-lingual detoxification without direct fine- tuning.

Copied to clipboard

Daryna Dementieva, Nikolay Babakov, Amit Ronen, Abinew Ali Ayele, Naquee Rizwan, Florian Schneider, Xintong Wang, Seid Muhie Yimam, Daniil Alekhseevich Moskovskiy, Elisei Stakovskii, Eran Kaufman, Ashraf Elnagar, Animesh Mukherjee, Alexander Panchenko

Cross-lingual Evidence Improves Monolingual Fake News Detection (2021.acl-srw)

Challenge:	Existing approaches to manage toxic speech on social platforms are limited . however, there is a need for more proactive moderation of abusive speech.
Approach:	They extend parallel text detoxification corpus to new languages to test the approach . they propose a method that combines toxic and non-toxic sentences into a more neutral form .
Outcome:	The proposed method integrates the descriptive features of toxic and non-toxic sentences into a more neutral or non- toxic form.

Copied to clipboard

Daryna Dementieva, Alexander Panchenko

Improving Neural Entity Disambiguation with Graph Embeddings (P19-2)

Challenge:	Existing methods focused on one language and do not use multilingual information.
Approach:	They propose a new technique based on cross-lingual evidence that can be used for fake news detection . they compared their proposed technique with strong baselines on two datasets of general-topic news .
Outcome:	The proposed technique improves existing methods and can be used on real and fake news datasets.

Copied to clipboard

Özge Sevgili, Alexander Panchenko, Chris Biemann

MERA: A Comprehensive LLM Evaluation in Russian (2024.acl-long)

Challenge:	Entity Disambiguation (ED) is the task of linking an ambiguous entity mention to a corresponding entry in a knowledge base.
Approach:	They propose a method that integrates structured information from the knowledge base with unstructured information from text-based representations.
Outcome:	The proposed method improves on a graph of hyperlinks between Wikipedia articles and a state-of-the-art neural ED model.

Copied to clipboard

Alena Fenogenova, Artem Chervyakov, Nikita Martynov, Anastasia Kozlova, Maria Tikhonova, Albina Akhmetgareeva, Anton Emelyanov, Denis Shevelev, Pavel Lebedev, Leonid Sinev, Ulyana Isaeva, Katerina Kolomeytseva, Daniil Moskovskiy, Elizaveta Goncharova, Nikita Savushkin, Polina Mikhailova, Anastasia Minaeva, Denis Dimitrov, Alexander Panchenko, Sergey Markov

Acceleration of Backpropagation in Linear Layers of Transformer Models Based on Gradient Structure (2026.eacl-srw)

Challenge:	Recent advances in foundation models have led to the emergence of powerful Large Language Models (LLMs), which showcase unprecedented tasksolving capabilities.
Approach:	They propose a method to evaluate FMs and LMs in fixed zero- and few-shot instruction settings that can be extended to other modalities.
Outcome:	The proposed evaluation methodology includes an open-source code base and a leaderboard with a submission system.

Copied to clipboard

Dmitrii Topchii, Alexander Panchenko, Viktoriia A. Chekalina

How Much Knowledge Can You Pack into a LoRA Adapter without Harming LLM? (2025.findings-naacl)

Challenge:	a sparsity-exploiting backward pass is a memory-efficient way to accelerate LLM fine-tuning.
Approach:	They propose a method that exploits padding-induced gradient sparsity to accelerate backward computation.
Outcome:	The proposed method achieves a backward pass speedup of 2.15x on GLUE and 1.99x on reasoning benchmarks while maintaining memory usage identical to the regular PyTorch fine-tuning.

Copied to clipboard

Sergey Pletenev, Maria Marina, Daniil Moskovskiy, Vasily Konovalov, Pavel Braslavski, Alexander Panchenko, Mikhail Salnikov

Every Child Should Have Parents: A Taxonomy Refinement Algorithm Based on Hyperbolic Term Embeddings (P19-1)

Challenge:	Low-rank adaptation (LoRA) is a popular training technique for updating or domain-specific adaptation of Large Language Models (LLMs).
Approach:	They propose to use low-rank adaptation to incorporate new facts into the LLM without compromising previously learned knowledge.
Outcome:	The proposed approach is harmful because the model's performance declines after such fine-tuning.

Copied to clipboard

Rami Aly, Shantanu Acharya, Alexander Ossa, Arne Köhn, Chris Biemann, Alexander Panchenko

Unsupervised Semantic Frame Induction using Triclustering (P18-2)

Challenge:	Existing approaches to domain-specific taxonomy induction from text are relying on distributional semantics for hyponym-hypernym relationships, but many of them learn prototypical hypernymes, not taking into account the relation between both terms in classification.
Approach:	They propose to use Poincaré embeddings to improve existing approaches to domain-specific taxonomy induction from text as a signal for relocating wrong hyponym terms and attaching disconnected terms in a taxonomies.
Outcome:	The proposed method significantly improves state-of-the-art methods on the SemEval-2016 Task 13 on taxonomy extraction.

Copied to clipboard

Dmitry Ustalov, Alexander Panchenko, Andrey Kutuzov, Chris Biemann, Simone Paolo Ponzetto

A System for Answering Simple Questions in Multiple Languages (2023.acl-demo)

Challenge:	Recent work on frame-semantics has enabled the development of wide-coverage frame parsers using supervised learning.
Approach:	They propose to use dependency triples to perform unsupervised frame induction on a Web-scale corpus.
Outcome:	The proposed approach performs state-of-the-art on a FrameNet-derived dataset and performs on par with competitive methods on . verb class clustering task.

Copied to clipboard

Anton Razzhigaev, Mikhail Salnikov, Valentin Malykh, Pavel Braslavski, Alexander Panchenko

Emergent Misalignment via In-Context Learning: Narrow in-context examples can produce broadly misaligned LLMs (2026.acl-long)

Challenge:	Existing knowledge graph question answering systems are limited to simple questions, but they can be used to answer complex questions.
Approach:	They propose a multilingual Knowledge Graph Question Answering technique that orders potential responses based on the distance between the question’s text embeddings and the answer’s graph embedds.
Outcome:	The proposed method consistently outperforms baseline systems, including seq2seq QA models and complex rule-based pipelines.

Copied to clipboard

Nikita Afonin, Nikita Andriianov, Vahagn Hovhannisyan, Nikhil Bageshpura, Kyle Liu, Kevin Zhu, Sunishchal Dev, Ashwinee Panda, Oleg Rogov, Elena Tutubalina, Alexander Panchenko, Mikhail Seleznyov

ParaDetox: Detoxification with Parallel Data (2022.acl-long)

Challenge:	Recent studies have documented emergent misalignment in language models adapted on narrow examples . Emergent misalignement occurs when models are trained on narrow set of misallocated examples resulting in harmful or misleading responses .
Approach:	They propose to explain in-context EM as conflict between safety objectives and context-following behavior.
Outcome:	The proposed model is adapted on 16 in-context examples and produces misaligned responses to benign queries.

Copied to clipboard

Varvara Logacheva, Daryna Dementieva, Sergey Ustyantsev, Daniil Moskovskiy, David Dale, Irina Krotova, Nikita Semenov, Alexander Panchenko

When Models Lie, We Learn: Multilingual Span-Level Hallucination Detection with PsiloQA (2025.findings-emnlp)

Challenge:	a pipeline for the collection of parallel data for the detoxification task is available.
Approach:	They propose a pipeline for the collection of parallel data for the detoxification task . they collect non-toxic paraphrases for over 10,000 English toxic sentences .
Outcome:	The proposed pipeline outperforms state-of-the-art models on both automatic and manual evaluations.

Copied to clipboard

Elisei Rykov, Kseniia Petrushina, Maksim Savkin, Valerii Olisov, Artem Vazhentsev, Kseniia Titova, Alexander Panchenko, Vasily Konovalov, Julia Belikova

When Punctuation Matters: A Large-Scale Comparison of Prompt Robustness Methods for LLMs (2025.findings-emnlp)

Challenge:	Existing hallucination detection benchmarks operate at the sequence level and are limited to English . Existing methods lacking fine-grained, multilingual supervision are limited in English based on the sequence .
Approach:	They propose a large-scale, multilingual dataset annotated with span-level hallucinations across 14 languages.
Outcome:	The proposed dataset annotated with span-level hallucinations across 14 languages is scalable and cost-efficient.

Copied to clipboard

Mikhail Seleznyov, Mikhail Chaichuk, Gleb Ershov, Alexander Panchenko, Elena Tutubalina, Oleg Somov

Building a Web-Scale Dependency-Parsed Corpus from CommonCrawl (L18-1)

Challenge:	Large Language Models (LLMs) are sensitive to subtle, non-semantic variations in prompt phrasing and formatting.
Approach:	They propose to evaluate 4 methods for improving prompt robustness within a unified experimental framework.
Outcome:	The proposed methods are compared to 8 models from Llama, Qwen and Gemma families and are generalized against multiple types of distribution shifts.

Copied to clipboard

Alexander Panchenko, Eugen Ruppert, Stefano Faralli, Simone P. Ponzetto, Chris Biemann

Challenge:	DepCC is the largest-to-date linguistically analyzed corpus in English . large corpora are essential for the modern data-driven approaches to natural language processing .
Approach:	They present a large-to-date linguistically analyzed corpus in English with 365 million documents . they build an index of all sentences and their linguistic meta-data enabling quick search across the corpus .
Outcome:	The proposed model outperforms state-of-the-art models on smaller corpora on the SimVerb3500 dataset.

A large-scale computational study of content preservation measures for text style transfer and paraphrase generation (2022.acl-srw)

Copied to clipboard

Nikolay Babakov, David Dale, Varvara Logacheva, Alexander Panchenko

Text Generation

Challenge:	Text style transfer and paraphrases generation are growing areas of NLP . many researchers still use BLEU-like measures to evaluate content preservation .
Approach:	They compare 57 different measures based on different principles on 19 annotated datasets . they find that measures relying on cross-encoder models outperform alternative approaches .
Outcome:	The proposed methods outperform traditional methods on 19 datasets.

An Unsupervised Word Sense Disambiguation System for Under-Resourced Languages (L18-1)

Copied to clipboard

Dmitry Ustalov, Denis Teslenko, Alexander Panchenko, Mikhail Chernoskutov, Chris Biemann, Simone Paolo Ponzetto

CAM 2.0: End-to-End Open Domain Comparative Question Answering System (2024.lrec-main)

Challenge:	Existing systems for word sense disambiguation are limited to the Russian language and lack of resources to address the problem.
Approach:	They propose an unsupervised system for word sense disambiguation that uses a traditional vector space model to estimate the most similar word sense corresponding to its context.
Outcome:	The proposed system outperforms the sparse mode on all datasets according to the adjusted Rand index.

Copied to clipboard

Ahmad Shallouf, Hanna Herasimchyk, Mikhail Salnikov, Rudy Alexandro Garrido Veliz, Natia Mestvirishvili, Alexander Panchenko, Chris Biemann, Irina Nikishina

Natural Language Interfaces

Challenge:	Comparative Question Answering is a Natural Language Processing task that combines Question Answers and Argument Mining.
Approach:	They propose a system for answering comparative questions called CAM 2.0 and a public leaderboard called CompUGE that unifies existing datasets under a single easy-to-use evaluation suite.
Outcome:	The proposed system is compared with previous web-form-based systems . it features question identification, object and aspect labeling, stance classification, summarization . the proposed system has a user-friendly interface and is available for free on the web .

xCOMET-lite: Bridging the Gap Between Efficiency and Quality in Learned MT Evaluation Metrics (2024.emnlp-main)

Copied to clipboard

Daniil Larionov, Mikhail Seleznyov, Vasiliy Viskov, Alexander Panchenko, Steffen Eger

Multimodal Evaluation of Russian-language Architectures (2026.eacl-long)

Challenge:	State-of-the-art trainable machine translation evaluation metrics rely on large encoders . this makes them computationally expensive and inaccessible to researchers with limited resources.
Approach:	They propose a method to extract knowledge stored in large encoders and a pipeline for efficient black-box distillation.
Outcome:	The proposed model surpasses COMET-22 and BLEURT-20 on the WMT22 dataset by 6.4%.

Copied to clipboard

Artem Chervyakov, Ulyana Isaeva, Anton Emelyanov, Artem Safin, Maria Tikhonova, Alexander Kharitonov, Yulia Lyakh, Petr Surovtsev, Denis Shevelev, Vildan Saburov, Vasily Konovalov, Elisei Rykov, Ivan Sviridov, Amina Miftakhova, Ilseyar Alimova, Alexander Panchenko, Alexander Kapitanov, Alena Fenogenova

Multimodality

Challenge:	Multimodal large language models (MLLMs) are at the center of research attention, yet intelligence, limitations, and risks remain insufficiently understood.
Approach:	They propose an open multimodal evaluation framework for Russian-spoken architectures . the framework is instruction-based and includes 18 newly constructed evaluation tasks .
Outcome:	The proposed framework provides a replicable methodology for constructing multimodal benchmarks in Russian-spoken architectures.

Word Sense Disambiguation for 158 Languages using Word Embeddings Only (2020.lrec-1)

Copied to clipboard

Varvara Logacheva, Denis Teslenko, Artem Shelmanov, Steffen Remus, Dmitry Ustalov, Andrey Kutuzov, Ekaterina Artemova, Chris Biemann, Simone Paolo Ponzetto, Alexander Panchenko