Tomasz Walkowiak
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Tomasz Walkowiak has a total of 62 co-authors. Here are the top co-authors by number of shared publications:
- Jacek Mazurkiewicz
Wrocław University of Science and Technology
20 shared publications - Dariusz Caban
Wrocław University of Science and Technology
14 shared publications - Henryk Maciejewski
Wrocław University of Science and Technology
9 shared publications - Mateusz Gniewkowski
Wrocław University of Science and Technology
7 shared publications - Kamil Szyc
Wrocław University of Science and Technology
5 shared publications Szymon Datko
Wrocław University of Science and Technology
5 shared publications
Latest Publications
- 7 September 2026
- Published by Springer Nature
- p. 75-92
Towards Personalized Language Models: Significant Advancements at the Cost of Emerging Risks
- 7 September 2026
- Published by Springer Nature
- p. 75-92
Towards Personalized Language Models: Significant Advancements at the Cost of Emerging Risks
- 12 June 2026
- Published by Springer Nature
- p. 468-477
Visualizing Authorial Style: How Chunk Size Affects UMAP and PaCMAP Representations
- 12 June 2026
- Published by Springer Nature
- p. 468-477
Visualizing Authorial Style: How Chunk Size Affects UMAP and PaCMAP Representations
- Open Access
- 16 December 2025
- Published by Uniwersytet Jagiellonski - Wydawnictwo Uniwersytetu Jagiellonskiego in Romanica Cracoviensia
- Vol. 25 (3) , 203
Au-delà de l’attribution d’auteur : la stylométrie permet-elle d’identifier l’« identité collective » des textes littéraires et le sexe des auteurs ? Analyse comparée de la fiction narrative d’Europe et d’Afrique
This article reports a stylometric study of contemporary French prose from the second half of the 20th century by authors from Europe and Francophone Africa. We test whether classical stylometry, long effective in authorship attribution, can also capture collective properties of large corpora – specifically author gender and cultural origin. Results show that, in fiction, gender detection is highly unreliable: skilled writers can convincingly adopt diverse voices regardless of sociological profile, whereas the same methods often succeed on non-fiction by users (blogs, emails, social media). Likewise, in French prose, analyses based on short lists of neutral function words fail to discriminate European from African authors. Clear separation appears only when long n-gram features are used, which encode culturally inflected vocabulary and contexts. We discuss methodological implications for stylometry beyond attribution.
- 16 December 2025
- Published by Uniwersytet Jagiellonski - Wydawnictwo Uniwersytetu Jagiellonskiego in Romanica Cracoviensia
- Vol. 25 (3) , 203
Open AccessAu-delà de l’attribution d’auteur : la stylométrie permet-elle d’identifier l’« identité collective » des textes littéraires et le sexe des auteurs ? Analyse comparée de la fiction narrative d’Europe et d’Afrique
This article reports a stylometric study of contemporary French prose from the second half of the 20th century by authors from Europe and Francophone Africa. We test whether classical stylometry, long effective in authorship attribution, can also capture collective properties of large corpora – specifically author gender and cultural origin. Results show that, in fiction, gender detection is highly unreliable: skilled writers can convincingly adopt diverse voices regardless of sociological profile, whereas the same methods often succeed on non-fiction by users (blogs, emails, social media). Likewise, in French prose, analyses based on short lists of neutral function words fail to discriminate European from African authors. Clear separation appears only when long n-gram features are used, which encode culturally inflected vocabulary and contexts. We discuss methodological implications for stylometry beyond attribution.
- 1 November 2025
- Published by Springer Nature
- p. 219-228
Classification of Literary Epochs by TF-IDF, Transformers, and Large Language Models
- 1 November 2025
- Published by Springer Nature
- p. 219-228
Classification of Literary Epochs by TF-IDF, Transformers, and Large Language Models
- 13 October 2025
- Published by John Benjamins Publishing Company
- p. 161-172
Linguistic correlates of semantic knowledge ontologies
This chapter provides a comparative analysis of bibliographic corpora including titles extracted from large national bibliographies (Czech, Finnish, German, Norwegian, and Polish). From the examined corpora, subsets were obtained, corresponding to the basic categories of the DDC/UDC formal ontologies (Dewey Decimal Classification and Universal Decimal Classification). The most relevant sets of keywords were then generated from these subsets and projected onto a common semantic space using automatic translation. The study revealed the existence of common ‘European’ sets of terms, corresponding to the large semantic domains defined in the DDC/UDC ontology.
- 13 October 2025
- Published by John Benjamins Publishing Company
- p. 161-172
Linguistic correlates of semantic knowledge ontologies
This chapter provides a comparative analysis of bibliographic corpora including titles extracted from large national bibliographies (Czech, Finnish, German, Norwegian, and Polish). From the examined corpora, subsets were obtained, corresponding to the basic categories of the DDC/UDC formal ontologies (Dewey Decimal Classification and Universal Decimal Classification). The most relevant sets of keywords were then generated from these subsets and projected onto a common semantic space using automatic translation. The study revealed the existence of common ‘European’ sets of terms, corresponding to the large semantic domains defined in the DDC/UDC ontology.
- Open Access
StylOch at PAN: Gradient-Boosted Trees with Frequency-Based Stylometric Features
This submission to the binary AI detection task is based on a modular stylometric pipeline, where: public spaCy models are used for text preprocessing (including tokenisation, named entity recognition, dependency parsing, part-of-speech tagging, and morphology annotation) and extracting several thousand features (frequencies of n-grams of the above linguistic annotations); light-gradient boosting machines are used as the classifier. We collect a large corpus of more than 500 000 machine-generated texts for the classifier's training. We explore several parameter options to increase the classifier's capacity and take advantage of that training set. Our approach follows the non-neural, computationally inexpensive but explainable approach found effective previously.
Open AccessStylOch at PAN: Gradient-Boosted Trees with Frequency-Based Stylometric Features
This submission to the binary AI detection task is based on a modular stylometric pipeline, where: public spaCy models are used for text preprocessing (including tokenisation, named entity recognition, dependency parsing, part-of-speech tagging, and morphology annotation) and extracting several thousand features (frequencies of n-grams of the above linguistic annotations); light-gradient boosting machines are used as the classifier. We collect a large corpus of more than 500 000 machine-generated texts for the classifier's training. We explore several parameter options to increase the classifier's capacity and take advantage of that training set. Our approach follows the non-neural, computationally inexpensive but explainable approach found effective previously.
- 30 June 2025
- Published by Springer Nature
- p. 49-60
Precise Language Deception: XAI Driven Targeted Adversarial Examples with Restricted Knowledge
- 30 June 2025
- Published by Springer Nature
- p. 49-60
Precise Language Deception: XAI Driven Targeted Adversarial Examples with Restricted Knowledge
- 27 May 2025
- Published by Springer Nature
- p. 221-228
Energy Efficiency in Large Language Models: An Empirical Study
- 27 May 2025
- Published by Springer Nature
- p. 221-228
Energy Efficiency in Large Language Models: An Empirical Study