While librarians have been describing books manually for decades, several libraries are now experimenting with methods to automatically create metadata.
Thanks to the growing volume of publications in electronic format, and the increase of artificial intelligence (AI) applications such as machine learning, there is a growing possibility for computers to interpret electronic texts and assist librarians when describing publications.
During a webinar on 8 November, Martijn Kleppe — Head of Research of the National Library of the Netherlands (KB) — discussed lessons learned by the KB learned during their first experiments, as recently described in a whitepaper.
Without going into technical details, he emphasized the lessons learned and gave advice to colleagues considering experimenting with AI techniques to automatically describe publications.
The webinar host was Jeannette Frey, President of LIBER and director of the Bibliothèque Cantonale et Universitaire (BCU) Lausanne. The slides have been published on Zenodo.