Implementing AI in the National Library of Spain: The ETSO Project and Stylometric Analysis of Golden Age Theatre
摘要
The Biblioteca Nacional de España (BNE), founded in 1711, serves as Spain’s national library, safeguarding one of the most significant collections of Spanish Golden Age theatre manuscripts. With a staff of 364 and an annual budget of €35.86 million, it is a cornerstone of Spain’s cultural preservation efforts. In 2017, the ETSO project (“Stylometry Applied to the Theater of the Golden Age”) was launched to address the lack of transcribed texts from this period. Using AI tools like Transkribus, ETSO has automated the transcription and modernisation of over 1800 theatrical works, achieving up to 97% accuracy for printed texts and 91% for manuscripts. The project is ongoing and continues to refine its methodologies while contributing to the CETSO corpus and the TEXORO search engine. The project faced challenges such as adapting AI to historical typefaces, modernising archaic spellings, and overcoming transcription errors in handwritten manuscripts. These were mitigated through extensive model training and interdisciplinary collaboration. ETSO demonstrates how AI can unlock cultural heritage, making historical texts accessible for research and public engagement and serving as a model for integrating technology into humanities projects.