par Van Hooland, Seth 
;Coeckelbergs, Mathias
Référence Handbook of Digital Public History, De Gruyter, page (517-529)
Publication Publié, 2022-04
;Coeckelbergs, Mathias Référence Handbook of Digital Public History, De Gruyter, page (517-529)
Publication Publié, 2022-04
Partie d'ouvrage collectif
| Résumé : | The current excitement in regards to machine learning has spurred enthusiasm amongst collection holders and historians alike to rely on algorithms to reduce the amount of manual labor required for management and appraisal of large volumes of non-structured archival content. The Digital Humanities and commercial archival software promote out-of-the-box tools for auto-classification, but is the adoption of machine learning as straightforward as it is currently presented in both the popular press and the Digital Humanities literature? This chapter brings a sense of pragmatism to the debate by giving an overview of both possibilities and limits of machine learning to extract semantics from large collections of digitized textual archives. Two methods have gained substantial popularity: Topic Modeling (TM) and Word Embeddings (WE). This chapter introduces these non-supervised machine learning methods to the community of historians, based on an experimental case-study of digitized archival holdings of the European Commission (EC). |
