Keywords: n-grams, textual borrowings, jaccard coefficient, tokenization, text normalization, quality metrics, f1-score
Analysis of the Impact of n-Gram Size on the Quality of Text Plagiarism Detection
UDC 004.912
This paper examines the impact of n-gram length on the quality of text plagiarism detection. The study employs an n-gram-based approach using the Jaccard coefficient to assess text similarity. Experiments were conducted on a corpus of scientific and educational texts containing various types of borrowed content, including verbatim, partial, and paraphrased instances. The results demonstrate that while low n-values increase detection recall, they reduce precision due to accidental matches. Conversely, increasing n causes the system to miss some actual instances of plagiarism. It is noted separately that the most balanced results are achieved in the range n = 3–4. In this case, an optimal balance between accuracy and completeness is observed. These findings can inform the configuration of simple plagiarism detection systems and educational anti-plagiarism modules.
1. Янников И.М. Методы и алгоритмы для поиска сходства между текстами / И.М. Янников, М.В. Ершова, А.Н. Исенбаев // Интеллектуальные системы в производстве. – 2024. – Т. 22, № 2. – С. 103–113.
2. Смирнов И.В. Интеллектуальный анализ текстов на основе методов разноуровневой обработки естественного языка / И.В. Смирнов. – М.: ФИЦ ИУ РАН, 2023. – 356 с.
3. Нугуманова А.Б. Автоматизированная обработка текстовых массивов: учебник и практикум для вузов / А.Б. Нугуманова. – 2-е изд. – М.: Издательство Юрайт, 2024. – 82 с.
4. Гасанов Э.Э. Интеллектуальные системы. Теория хранения и поиска информации: учебник для вузов / Э.Э. Гасанов, В.Б. Кудрявцев. – 2-е изд., испр. и доп. – М.: Издательство Юрайт, 2025. – 271 с.
Keywords: n-grams, textual borrowings, jaccard coefficient, tokenization, text normalization, quality metrics, f1-score
For citation: Nazarov A.A. , Preobrazhenskiy A.P. , Analysis of the Impact of n-Gram Size on the Quality of Text Plagiarism Detection. Bulletin of the Voronezh Institute of High Technologies. 2026;20(3). Available from: https://vestnikvivt.ru/ru/journal/pdf?id=1505 (In Russ).
Received 02.07.2026
Revised 12.08.2026
Accepted 12.08.2026