The article investigates the continued importance of keywords in digital humanities and especially their relation to recent machine-learning approaches. Different research practices related to digital humanities agree on the importance of keywords to present issues and/or provide the baseline for new stories about the past, about literature or media. Keywords are useful to target ideas that have no clear definitions and productive in describing contested categorisations that are key to humanities scholarship. At the same time, keywords are often employed used without considering specific contexts and how they are generated. To understand the diversity of keywords in digital humanities, we consider three approaches to computationally generating keywords, from traditional and established ones to state-of-the-art language modelling. With these three approaches, we analyse a case where keywords should be especially powerful, as underlying considerations are uncertain. We cover the relation between security, human-rights and freedom according to discussions in United Nations documents. Finally, we present a number of approaches to productively use keywords to tell a different story about the relation of security and freedom according to UN discussions.
Blanke et al. (Fri,) studied this question.