In EDS ansehen

Contextualizing ancient texts with generative neural networks.

Gespeichert in:
Bibliographische Detailangaben
Titel: Contextualizing ancient texts with generative neural networks.
Autoren: Assael Y; Google DeepMind, London, UK. yannisassael@google.com., Sommerschield T; Department of Classics and Archaeology, University of Nottingham, Nottingham, UK. thea.sommerschield@durham.ac.uk., Cooley A; Department of Classics and Ancient History, University of Warwick, Warwick, UK., Shillingford B; Google DeepMind, London, UK., Pavlopoulos J; Department of Informatics, Athens University of Economics and Business, Athens, Greece., Suresh P; Google DeepMind, London, UK., Herms B; Google, Mountain View, CA, USA., Grayston J; Google, Mountain View, CA, USA., Maynard B; Google, Mountain View, CA, USA., Dietrich N; Google DeepMind, London, UK., Wulgaert R; Sint-Lievenscollege, Ghent, Belgium., Prag J; Faculty of Classics, University of Oxford, Oxford, UK., Mullen A; Department of Classics and Archaeology, University of Nottingham, Nottingham, UK., Mohamed S; Google DeepMind, London, UK.
Quelle: Nature [Nature] 2025 Sep; Vol. 645 (8079), pp. 141-147. Date of Electronic Publication: 2025 Jul 23.
Publikationsart: Journal Article; Historical Article
Sprache: English
Info zur Zeitschrift: Publisher: Nature Publishing Group Country of Publication: England NLM ID: 0410462 Publication Model: Print-Electronic Cited Medium: Internet ISSN: 1476-4687 (Electronic) Linking ISSN: 00280836 NLM ISO Abbreviation: Nature Subsets: MEDLINE
Imprint Name(s): Publication: Basingstoke : Nature Publishing Group
Original Publication: London, Macmillan Journals ltd.
MeSH-Schlagworte: Neural Networks, Computer*, History, Ancient ; Humans ; Civilization/history
Abstract: Competing Interests: Competing interests: The authors declare no competing interests.
Human history is born in writing. Inscriptions are among the earliest written forms, and offer direct insights into the thought, language and history of ancient civilizations. Historians capture these insights by identifying parallels-inscriptions with shared phrasing, function or cultural setting-to enable the contextualization of texts within broader historical frameworks, and perform key tasks such as restoration and geographical or chronological attribution 1 . However, current digital methods are restricted to literal matches and narrow historical scopes. Here we introduce Aeneas, a generative neural network for contextualizing ancient texts. Aeneas retrieves textual and contextual parallels, leverages visual inputs, handles arbitrary-length text restoration, and advances the state of the art in key tasks. To evaluate its impact, we conduct a large study with historians using outputs from Aeneas as research starting points. The historians find the parallels retrieved by Aeneas to be useful research starting points in 90% of cases, improving their confidence in key tasks by 44%. Restoration and geographical attribution tasks yielded superior results when historians were paired with Aeneas, outperforming both humans and artificial intelligence alone. For dating, Aeneas achieved a 13-year distance from ground-truth ranges. We demonstrate Aeneas' contribution to historical workflows through analysis of key traits in the renowned Roman inscription Res Gestae Divi Augusti, showing how integrating science and humanities can create transformative tools to assist historians and advance our understanding of the past.
(© 2025. The Author(s).)
References: Proc Natl Acad Sci U S A. 2016 Apr 26;113(17):4664-9. (PMID: 27071103)
Sci Rep. 2023 Oct 6;13(1):16887. (PMID: 37803108)
Proc Natl Acad Sci U S A. 2020 Sep 15;117(37):22743-22751. (PMID: 32873650)
Proc Natl Acad Sci U S A. 2009 Aug 18;106(33):13685-90. (PMID: 19666571)
PLoS One. 2019 May 8;14(5):e0215775. (PMID: 31067260)
Nature. 2022 Mar;603(7900):280-283. (PMID: 35264762)
PLoS One. 2022 Jun 16;17(6):e0269869. (PMID: 35709226)
PLoS One. 2010 Mar 19;5(3):e9506. (PMID: 20333254)
Entry Date(s): Date Created: 20250723 Date Completed: 20250903 Latest Revision: 20250906
Update Code: 20250906
PubMed Central ID: PMC12408360
DOI: 10.1038/s41586-025-09292-5
PMID: 40702185
Datenbank: MEDLINE
Beschreibung
Abstract:Competing Interests: Competing interests: The authors declare no competing interests.<br />Human history is born in writing. Inscriptions are among the earliest written forms, and offer direct insights into the thought, language and history of ancient civilizations. Historians capture these insights by identifying parallels-inscriptions with shared phrasing, function or cultural setting-to enable the contextualization of texts within broader historical frameworks, and perform key tasks such as restoration and geographical or chronological attribution <sup>1</sup> . However, current digital methods are restricted to literal matches and narrow historical scopes. Here we introduce Aeneas, a generative neural network for contextualizing ancient texts. Aeneas retrieves textual and contextual parallels, leverages visual inputs, handles arbitrary-length text restoration, and advances the state of the art in key tasks. To evaluate its impact, we conduct a large study with historians using outputs from Aeneas as research starting points. The historians find the parallels retrieved by Aeneas to be useful research starting points in 90% of cases, improving their confidence in key tasks by 44%. Restoration and geographical attribution tasks yielded superior results when historians were paired with Aeneas, outperforming both humans and artificial intelligence alone. For dating, Aeneas achieved a 13-year distance from ground-truth ranges. We demonstrate Aeneas' contribution to historical workflows through analysis of key traits in the renowned Roman inscription Res Gestae Divi Augusti, showing how integrating science and humanities can create transformative tools to assist historians and advance our understanding of the past.<br /> (© 2025. The Author(s).)
ISSN:1476-4687
DOI:10.1038/s41586-025-09292-5