EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Attributing authorship via the perplexity of authorial language models

Weihang Huang iD, Akira Murakami, Jack Grieve iD

PLOS One · 03.07.2025 · 20 (7) · e0327081

DOI: 10.1371/journal.pone.0327081 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PLOS One, online erschienen am 03.07.2025. Verfasst von Weihang Huang und 2 weiteren Autor:innen. Beteiligte Einrichtung: Department of Linguistics and Communication, University of Birmingham, Birmingham, West Midlands, United Kingdom. Hauptthemen laut MEDLINE: Authorship, Language, Models, Theoretical. Die Arbeit wurde bisher 5-mal zitiert (Crossref) und nennt 48 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Office of the Director of National Intelligence (ODNI), Intelligence Advanced Research Projects Activity.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Authorship attribution is the task of identifying the most likely author of a questioned document from a set of candidate authors, where each candidate is represented by a writing sample. A wide range of quantitative methods for inferring authorship have been developed in stylometry, but the rise of Large Language Models (LLMs) offers new opportunities in this field. In this paper, we introduce a technique for authorship attribution based on fine-tuned LLMs. Our approach involves first further pretraining LLMs for each candidate author based on their known writings and then assigning the questioned document to the author whose Authorial Language Model (ALM) finds the questioned document most predictable, measured as the perplexity of the questioned document. We find that our approach meets or exceeds the current state-of-the-art on several standard benchmarking datasets. In addition, we show how our approach can be used to measure the predictability of each word in a questioned document for a given candidate ALM, allowing the linguistic patterns that drive our attributions to be inspected directly. Finally, we analyze what types of words generally drive successful attributions, finding that content words classes are characterized by a higher density of authorship information than function word classes, challenging a long-standing assumption of stylometry.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

AuthorshipLanguageModels, TheoreticalHumansLinguistics

Autor:innen und Institutionen

  • Department of Linguistics and Communication, University of Birmingham, Birmingham, West Midlands, United Kingdom.

ORCID-Profile: Weihang Huang, Jack Grieve

Förderung

  • Office of the Director of National Intelligence (ODNI), Intelligence Advanced Research Projects Activity

Zitieren

APA

Huang, W., Murakami, A., & Grieve, J. (2025). Attributing authorship via the perplexity of authorial language models. PLOS One, 20(7), e0327081. https://doi.org/10.1371/journal.pone.0327081

Vancouver

Huang W, Murakami A, Grieve J. Attributing authorship via the perplexity of authorial language models. PLOS One. 2025;20(7):e0327081. doi:10.1371/journal.pone.0327081

BibTeX
@article{Huang2025,
  author = {Huang, Weihang and Murakami, Akira and Grieve, Jack},
  title = {Attributing authorship via the perplexity of authorial language models},
  journal = {PLOS One},
  year = {2025},
  volume = {20},
  number = {7},
  pages = {e0327081},
  doi = {10.1371/journal.pone.0327081}
}
Kostenlos starten →