PolEval 2021 papers and presentations
The following PolEval 2021 submissions were presented at the 4th AI & NLP Conference (online, October 25, 2021):
PolEval 2021: Introduction – Maciej Ogrodniczuk, Łukasz Kobyliński
Task 1: Punctuation restoration from read text
- PolEval 2021 Task 1: Punctuation Restoration from Read Text – Agnieszka Mikołajczyk, Piotr Pęzik, Adam Wawrzyński, Adam Kaczmarek, Wojciech Janowski, Michał Adamczyk
- Punctuation Restoration with Transformers – Krzysztof Wróbel, Dmytro Zhylko
- Punctuation Restoration with Ensemble of Neural Network Classifier and Pre-trained Transformers – Michał Marcińczuk
- Punctuation Restoration from Read Text with Transformer-based Tagger – Tomasz Ziętkiewicz
Task 2: Evaluation of translation quality assessment metrics
- Simple Recipes for Assessing Translation Quality – Dariusz Kłeczek
- Transformer as Machine Translation Evaluation Metrics – Krzysztof Wróbel
Task 3: Post-correction of OCR results
- PolEval 2021 Task 3: Post-correction of OCR Results – Szymon Rynkun, Łukasz Kobyliński, Witold Kieraś
- Post-correction of OCR Results Using Pre-trained Language Model – Mateusz Piotrowski
- OCR Correction with Encoder-Decoder Transformer – Krzysztof Wróbel
- Simple Yet Effective Method of Post-correcting OCR Errors – Paweł Dyda
- OCR Post-Correction with Heuristics – Michał Marcińczuk
Task 4: Question answering challenge
- PolEval 2021 Task 4: Question Answering Challenge – Maciej Ogrodniczuk, Piotr Przybyła
- Search Augmented Question Answering System Using Multilangual Transformer Model – Mateusz Piotrowski
- Answering Polish Trivia Questions with the Help of Dense Passage Retriever – Aleksander Smywiński-Pohl, Dmytro Zhylko, Krzysztof Wróbel, Magda Król
- Retrieve and Refine System for Polish Question Answering – Piotr Rybak
- Simple Recipes for Question Answering – Dariusz Kłeczek
Workshop proceedings were published by the Institute of Computer Science, Polish Academy of Sciences. Please cite the following:
@proceedings{ogr:kob:21:poleval,
editor = {Maciej Ogrodniczuk and Łukasz Kobyliński},
title = {{Proceedings of the PolEval 2021 Workshop}},
year = {2021},
address = {Warsaw, Poland},
publisher = {Institute of Computer Science, Polish Academy of Sciences},
url = {http://poleval.pl/files/poleval2021.pdf},
isbn = "978-83-63159-31-3"}
}
The winners of each of the tasks were also invited to submit an extended version of their papers to a Special Issue entitled “Creative and Generative Natural Language Processing and Its Applications” in an open access journal “Electronics” (ISSN 2079-9292; CODEN: ELECGJ, IF 2.412).