Authorship Verification Using Cloze Test with Large Language
Models

D 2025

Authorship Verification Using Cloze Test with Large Language Models

FOLTÝNEK, Tomáš; Tomáš KANCKO a Pavel RYCHLÝ

Základní údaje

Originální název

Authorship Verification Using Cloze Test with Large Language Models

Autoři

FOLTÝNEK, Tomáš; Tomáš KANCKO a Pavel RYCHLÝ

Vydání

Shoumen, Bulgaria, Proceedings of Recent Advances in Natural Language Processing, od s. 369-377, 9 s. 2025

Nakladatel

INCOMA Ltd.

Další údaje

Jazyk

angličtina

Typ výsledku

Stať ve sborníku

Obor

10200 1.2 Computer and information sciences

Stát vydavatele

Bulharsko

Utajení

není předmětem státního či obchodního tajemství

Forma vydání

elektronická verze "online"

Odkazy

URL

Označené pro přenos do RIV

Ano

Kód RIV

RIV/00216224:14330/25:00144190

Organizační jednotka

Fakulta informatiky

ISBN

978-954-452-098-4

ISSN

DOI

https://doi.org/10.26615/978-954-452-098-4-045

EID Scopus

2-s2.0-105034113497

Klíčová slova anglicky

Authorship verification; Cloze test; LLM

Příznaky

Mezinárodní význam, Recenzováno

Změněno: 9. 4. 2026 11:54, RNDr. Pavel Šmerk, Ph.D.

Anotace

V originále

Assignment outsourcing, also known as contract cheating, occurs when a student outsources an assessment task or a part of it to a third party. It has been one of the most pressing ethical issues in university education and was further exacerbated by the wide availability of chatbots based on large language models. We propose a method that has the potential to verify the authorship of a document in question by f illing in a cloze test. A close test with 10 items selected by our method can be used as a classif ier with an accuracy of 0.988 and a F1 score of 0.937. We also describe a general method for building a cloze-test-based classifier when the probability of authors and non-authors correctly filling in cloze items is known.

Návaznosti

TQ01000110, projekt VaV

Název: Webová aplikace na prokazování autorství

Investor: Technologická agentura ČR, Webová aplikace na prokazování autorství, DC3 - SHUV

Citovat

FOLTÝNEK, Tomáš; Tomáš KANCKO a Pavel RYCHLÝ. Authorship Verification Using Cloze Test with Large Language Models. Online. In Galia Angelova, Maria Kunilovskaya, Marie Escribe, Ruslan Mitkov. Proceedings of Recent Advances in Natural Language Processing. Shoumen, Bulgaria: INCOMA Ltd., 2025, s. 369-377. ISBN 978-954-452-098-4. Dostupné z: https://doi.org/10.26615/978-954-452-098-4-045.

@inproceedings{2561717,
   author = {Foltýnek, Tomáš and Kancko, Tomáš and Rychlý, Pavel},
   address = {Shoumen, Bulgaria},
   booktitle = {Proceedings of Recent Advances in Natural Language Processing},
   doi = {https://doi.org/10.26615/978-954-452-098-4-045},
   editor = {Galia Angelova, Maria Kunilovskaya, Marie Escribe, Ruslan Mitkov},
   keywords = {Authorship verification; Cloze test; LLM},
   howpublished = {elektronická verze "online"},
   language = {eng},
   location = {Shoumen, Bulgaria},
   isbn = {978-954-452-098-4},
   pages = {369-377},
   publisher = {INCOMA Ltd.},
   title = {Authorship Verification Using Cloze Test with Large Language Models},
   url = {https://aclanthology.org/2025.ranlp-1.45/},
   year = {2025}
}

TY  - CONF
ID  - 2561717
AU  - Foltýnek, Tomáš - Kancko, Tomáš - Rychlý, Pavel
PY  - 2025
TI  - Authorship Verification Using Cloze Test with Large Language Models
PB  - INCOMA Ltd.
CY  - Shoumen, Bulgaria
SN  - 9789544520984
KW  - Authorship verification
KW  - Cloze test
KW  - LLM
UR  - https://aclanthology.org/2025.ranlp-1.45/
N2  - Assignment outsourcing, also known as contract cheating, occurs when a student outsources an assessment task or a part of it to a third party. It has been one of the most pressing ethical issues in university education and was further exacerbated by the wide availability of chatbots based on large language models. We propose a method that has the potential to verify the authorship of a document in question by f illing in a cloze test. A close test with 10 items selected by our method can be used as a classif ier with an accuracy of 0.988 and a F1 score of 0.937. We also describe a general method for building a cloze-test-based classifier when the probability of authors and non-authors correctly filling in cloze items is known.
ER  -

FOLTÝNEK, Tomáš; Tomáš KANCKO a Pavel RYCHLÝ. Authorship Verification Using Cloze Test with Large Language Models. Online. In Galia Angelova, Maria Kunilovskaya, Marie Escribe, Ruslan Mitkov. \textit{Proceedings of Recent Advances in Natural Language Processing}. Shoumen, Bulgaria: INCOMA Ltd., 2025, s.~369-377. ISBN~978-954-452-098-4. Dostupné z: https://doi.org/10.26615/978-954-452-098-4-045.

Přehled o publikaci