Recent advances in Artificial Intelligence and Machine Learning (ML) have facilitated unprecedented analyses of ancient languages with unparalleled detail. ML methods now enable the exploration of several tasks associated with the study of ancient texts, including decipherment, digitization, attribution, studies in intertextuality and semantics.
This talk will provide an overview of key tasks, trends, and transformations within the multifaceted realm of ML for Ancient Languages, drawing inspiration from my personal experiences in this field. We will trace the evolution of the various ML techniques available to historical researchers, highlighting significant contributions and addressing persistent challenges.
Our starting focus will be on the effective impact of ML methods on the study of ancient languages, examining recent groundbreaking work across tasks, languages, scripts and media (from Cuneiform to the Indus script, from Oracle Bone inscriptions to carbonised papyri), employing diverse ML methods. Emphasis will be placed on the influence of conferences and competitions, the importance of FAIR datasets and interdisciplinary collaborations, and the development of explainable AI tools.
Our attention will then shift to extant challenges in the field, including ancient data bias, uneven standards, benchmarks and metrics, the risk of digital colonialism, and the unique difficulties posed by ancient texts. The discussion will close with a consideration of how emerging advancements in Generative AI may be responsibly harnessed for future progress.
Vortrag im Rahmen des „Digital History“-Forschungskolloquiums
Zeit: Mittwoch, 7. Februar 2024, 16-18 Uhr
Ort: Zoom-Konferenz (Zugang auf Anfrage oder via Mailingliste)
Bildnachweise
Beitragsbild: Assael et al. in Nature 603, 280–283 (2022) Fig. 1: inscription IG I3 4B restored.
Abb. 1: Sommerschield et al. in Computational Linguistics 49 (3), 703–747 (2023).
OpenEdition schlägt Ihnen vor, diesen Beitrag wie folgt zu zitieren:
Digital History Berlin (Redaktion) (2. Februar 2024). Thea Sommerschield: Machine Learning for Ancient Languages. Trends, challenges, and future prospects for a digital ancient history. Digital History Berlin. Abgerufen am 15. Oktober 2024 von https://doi.org/10.58079/vqpa