Dating Greek Papyri with Text Regression

Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics Volume 1: Long Papers, pages 10001–10013 July 9-14, 2023 ©2023 Association for Computational Linguistics Dating Greek Papyri with Text Regression John Pavlopoulos♠♣, Maria Konstantinidou⋄, Isabelle Marthot-Santaniello◦, Holger Essler‡, Asimina Paparigopoulou♠ ♠Department of Informatics, Athens University of Economics and Business, Greece {annis,asimina}@aueb.gr ♣Department of Computer and Systems Sciences, Stockholm University, Sweden ⋄ Democritus University of Thrace mkonst@helit.duth.gr ◦ University of Basel, Switzerland i.marthot-santaniello@unibas.ch ‡ Ca’Foscari University of Venice, Italy holger.essler@unive.it Abstract Dating Greek papyri accurately is crucial not only to edit their texts but also to understand numerous other aspects of ancient writing, doc- ument and book production and circulation, as well as various other aspects of administration, everyday life and intellectual history of antiq- uity. Although a substantial number of Greek papyri documents bear a date or other conclu- sive data as to their chronological placement, an even larger number can only be dated tenta- tively or in approximation, due to the lack of decisive evidence. By creating a dataset of 389 transcriptions of documentary Greek papyri, we train 389 regression models and we predict a date for the papyri with an average MAE of 54 years and an MSE of 1.17, outperforming image classifiers and other baselines. Last, we release date estimations for 159 manuscripts, for which only the upper limit is known. 1 Introduction Ancient textual artefacts are arguably the richest source of information on the ancient world. In the Graeco-Roman world and particularly in its Greek- speaking part, the most extensive coeval texts come from inscriptions and papyri. The latter is a col- lective term used for all ancient manuscripts, re- gardless of their writing material which, apart from papyrus, may be parchment, pottery, wood, and others. To correctly evaluate and make good use of these texts, we need to determine their date, prove- nance and historical context of their production and use. As far as dating is concerned, the value of the relevant evidence provided by the artefacts them- selves varies considerably, ranging from a direct date in the text (following, of course, the calendar and dating system of the respective historical pe- riod) to no evidence at all. In between, there are Figure 1: Papyrus ‘P. Basel 2 15’. Credit: University of Basel. https://papyri.info/ddbdp/p.bas;2;15 texts containing references to known historical fig- ures and events of a certain period, papyri which have been found next to other objects that can be dated, or other indirect evidence. The presence or absence of a date depends on the type of text preserved on the papyrus and its use through time, as well as on its state of conservation. Just like in modern times, it is much more likely to include a date in an official letter than in a page torn from a novel book. At the same time, it is more probable to find a date in a fully surviving letter than in a damaged one missing, for instance, the upper part of the first …

Content truncated to 500 words