Papers by Ondrej Dusek
LEEETs-Dial: Linguistic Entrainment in End-to-End Task-oriented Dialogue systems (2024.findings-naacl)
Copied to clipboard
| Challenge: | Linguistic entrainment is a phenomenon where linguistic patterns employed by conversational participants converge to one another. |
| Approach: | They propose methods for achieving dialogue entrainment in a task-oriented dialogue system using shared vocabulary. |
| Outcome: | The proposed model produces significantly better entrainment than the base model. |
Critic-Driven Decoding for Mitigating Hallucinations in Data-to-text Generation (2023.emnlp-main)
Copied to clipboard
| Challenge: | Hallucination of text lacking grounding in input data is a problem in neural data-to-text generation. |
| Approach: | They propose to combine probabilistic output of a generator language model with the output of an “text critic” classifier which guides the generation by assessing the match between the input data and the generated text. |
| Outcome: | The proposed method improves on the WebNLG and OpenDialKG benchmarks. |
Modular Monolingual Adaptation using Pretrained Language Models (2026.acl-industry)
Copied to clipboard
| Challenge: | Existing approaches to building monolingual models for low-resource languages require a full model tuning process. |
| Approach: | They propose a modular approach to build monolingual models for low-resource languages by finetuning the whole model on the target language. |
| Outcome: | The proposed model improves on natural language understanding tasks on Scottish Gaelic, Irish, and Quechua with Quechuan being a very low-resource language. |
SRS-Stories: Vocabulary-constrained multilingual story generation for language learning (2025.emnlp-industry)
Copied to clipboard
| Challenge: | Existing methods for learning foreign languages are to use a spaced repetition system to learn new vocabulary. |
| Approach: | They use large language models to generate personalized stories using only the vocabulary they know. |
| Outcome: | The generated stories are more grammatical, coherent, and provide better examples of word usage than the standard beam search approach. |
Neural Pipeline for Zero-Shot Data-to-Text Generation (2022.acl-long)
Copied to clipboard
| Challenge: | In data-to-text generation, training on in-domain data leads to overfitting and repeating training data noise. |
| Approach: | They propose to train pretrained language models on general-domain text-based operations by transforming single-item descriptions with modules trained on ordering, aggregation, and paragraph compression. |
| Outcome: | The proposed approach enables D2T generation from RDF triples in zero-shot settings. |
Reasoning Gets Harder for LLMs Inside A Dialogue (2026.acl-long)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) achieve strong performance on many reasoning benchmarks, yet these evaluations typically focus on isolated tasks that differ from real-world usage in task-oriented dialogue (TOD). |
| Approach: | They propose to use a dynamic benchmark to examine how framing reasoning tasks within task-oriented dialogue (TOD) affect LLM performance. |
| Outcome: | The proposed model performs well on isolated tasks and in task-oriented dialogues, but performance is inconsistent between them. |
TabGenie: A Toolkit for Table-to-Text Generation (2023.acl-demo)
Copied to clipboard
| Challenge: | TabGenie enables researchers to explore, preprocess, and analyze data-to-text generation datasets. |
| Approach: | They present TabGenie, a toolkit which enables researchers to explore, preprocess, and analyze a variety of data-to-text generation datasets. |
| Outcome: | The toolkit provides an interactive mode for debugging table-to-text generation, side-by-side comparison of generated system outputs, and easy exports for manual analysis. |
Mind the Labels: Describing Relations in Knowledge Graphs With Pretrained Models (2023.eacl-main)
Copied to clipboard
| Challenge: | Pretrained language models (PLMs) for data-to-text generation produce inaccurate outputs if labels are ambiguous or incomplete, which is often the case in D2T datasets. |
| Approach: | They propose to use a dataset to descib a relation between two entities using relation labels to train pretrained language models. |
| Outcome: | The proposed models are robust to generalizing to out-of-domain domains on a dataset for descibing a relation between two entities. |
Beyond Traditional Benchmarks: Analyzing Behaviors of Open LLMs on Data-to-Text Generation (2024.acl-long)
Copied to clipboard
| Challenge: | Existing benchmarks for data-to-text generation are saturated, and there is no way to test them. |
| Approach: | They propose a tool for collecting structured data from public APIs to analyze the behavior of open large language models on the task of data-to-text generation. |
| Outcome: | The proposed model can generate fluent and coherent texts in zero-shot settings from data in common formats collected with Quintd. |
GEMv2: Multilingual NLG Benchmarking in a Single Line of Code (2022.emnlp-demos)
Copied to clipboard
Sebastian Gehrmann, Abhik Bhattacharjee, Abinaya Mahendiran, Alex Wang, Alexandros Papangelis, Aman Madaan, Angelina Mcmillan-major, Anna Shvets, Ashish Upadhyay, Bernd Bohnet, Bingsheng Yao, Bryan Wilie, Chandra Bhagavatula, Chaobin You, Craig Thomson, Cristina Garbacea, Dakuo Wang, Daniel Deutsch, Deyi Xiong, Di Jin, Dimitra Gkatzia, Dragomir Radev, Elizabeth Clark, Esin Durmus, Faisal Ladhak, Filip Ginter, Genta Indra Winata, Hendrik Strobelt, Hiroaki Hayashi, Jekaterina Novikova, Jenna Kanerva, Jenny Chim, Jiawei Zhou, Jordan Clive, Joshua Maynez, João Sedoc, Juraj Juraska, Kaustubh Dhole, Khyathi Raghavi Chandu, Laura Perez Beltrachini, Leonardo F . R. Ribeiro, Lewis Tunstall, Li Zhang, Mahim Pushkarna, Mathias Creutz, Michael White, Mihir Sanjay Kale, Moussa Kamal Eddine, Nico Daheim, Nishant Subramani, Ondrej Dusek, Paul Pu Liang, Pawan Sasanka Ammanamanchi, Qi Zhu, Ratish Puduppully, Reno Kriz, Rifat Shahriyar, Ronald Cardenas, Saad Mahamood, Salomey Osei, Samuel Cahyawijaya, Sanja Štajner, Sebastien Montella, Shailza Jolly, Simon Mille, Tahmid Hasan, Tianhao Shen, Tosin Adewumi, Vikas Raunak, Vipul Raheja, Vitaly Nikolaev, Vivian Tsai, Yacine Jernite, Ying Xu, Yisi Sang, Yixin Liu, Yufang Hou
| Challenge: | Evaluations in machine learning rarely use the latest metrics, datasets, or human evaluation in favor of remaining compatible with prior work. |
| Approach: | They propose to use the Generation, Evaluation, and Metrics Benchmark to integrate new evaluation methods into existing evaluations. |
| Outcome: | The proposed evaluation infrastructure bridges the gap between the advantages of leaderboards and in-depth and evolving evaluations by allowing model developers to benefit from each other's work. |
Real-World Summarization: When Evaluation Reaches Its Limits (2025.findings-emnlp)
Copied to clipboard
| Challenge: | a recent study examines the evaluation of hotel highlights in the context of hotel data. |
| Approach: | They examine evaluation of faithfulness to input data in the context of hotel highlights . they compare traditional metrics, trainable methods, and LLM-as-a-judge approaches . |
| Outcome: | The results show that simple metrics outperform human judgments on LLM-generated summaries . the results also highlight challenges in crowdsourced evaluations. |
Can Large Language Models Personalize Dialogues to Generational Styles? (2025.findings-emnlp)
Copied to clipboard
| Challenge: | a human evaluation reveals that annotators were able to most accurately identify the generation behind P-MultiWoZ dialogues, based only on a single query-reply pair. |
| Approach: | They create a personalized, generation-specific version of MultiWOZ 2.2 by prompting LLMs to generate personalized dialogue responses. |
| Outcome: | The proposed model is a personalized version of MultiWOZ 2.2 for Generation X, Y, and Z . it is validated by automatic and human evaluations to determine whether it reflects generational linguistic traits. |
One-step Nonautoregressive Natural Language Generation with Shortcut Flow Matching Models (2026.acl-short)
Copied to clipboard
| Challenge: | Recent diffusion-based approaches to text generation are inefficient due to the need for multiple denoising steps. |
| Approach: | They propose a shortcut flow-matching model that learns to directly predict multi-step denoising outcomes in a single step. |
| Outcome: | The proposed model improves on three datasets and can predict multi-step denoising outcomes in a single step. |
Leak, Cheat, Repeat: Data Contamination and Evaluation Malpractices in Closed-Source LLMs (2024.eacl-long)
Copied to clipboard
| Challenge: | Lack of access to model details has raised concerns about data contamination among researchers. |
| Approach: | They conduct the first systematic analysis of work using OpenAI’s GPT-3.5 and GPT-4, the most prominently used LLMs today, in the context of data contamination. |
| Outcome: | The proposed models have been exposed to 4.7M samples from 263 benchmarks during the first year after their release. |
Faithful and Plausible Natural Language Explanations for Image Classification: A Pipeline Approach (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Existing explanation methods for image classification struggle to provide faithful and plausible explanations for predictions. |
| Approach: | They propose a natural language explanation method that can be applied to any CNN-based classifier without altering its training process or affecting predictive performance. |
| Outcome: | The proposed method can be applied to any CNN-based classifier without altering its training process or affecting predictive performance. |
Evaluating Text Style Transfer Evaluation: Are There Any Reliable Metrics? (2025.naacl-srw)
Copied to clipboard
| Challenge: | Text style transfer (TST) is a multidimensional task requiring the assessment of style transfer accuracy, content preservation, and naturalness. |
| Approach: | They propose to use text style transfer metrics to evaluate outputs of text editors . they also investigate the potential of large language models as tools for TST evaluation . |
| Outcome: | The proposed methods provide better insights than existing metrics, the authors show . their meta-evaluation through correlation with hu-man judgments shows they are effective . |
LLM Agents Implement an NLG System from Scratch: Building Interpretable Rule-Based RDF-to-Text Generators (2025.emnlp-industry)
Copied to clipboard
| Challenge: | Existing neural approaches to generate RDF-to-text are limited in their implementation. |
| Approach: | They propose a framework where the model is “trained” through collaborative interactions among multiple LLM agents rather than traditional backpropagation. |
| Outcome: | The proposed framework reduces hallucinations and fluency penalties on the WebNLG and OpenDialKG datasets. |
Ask the experts: sourcing a high-quality nutrition counseling dataset through Human-AI collaboration (2024.findings-emnlp)
Copied to clipboard
Simone Balloccu, Ehud Reiter, Karen Li, Rafael Sargsyan, Vivek Kumar, Diego Reforgiato, Daniele Riboni, Ondrej Dusek
| Challenge: | Large Language Models (LLMs) are being used by end-users for various tasks, including sensitive ones such as health counseling, disregarding potential safety concerns. |
| Approach: | They use ChatGPT to crowd-source dietary struggles and work with nutrition experts to generate supportive text using ChatGPS. |
| Outcome: | The proposed model outperforms other models on dietary struggles and mental health tasks. |