We address a three-fold challenges in data science for Computational Social Science and Digital Humanities: there is too much information for humans to analyse, it is too complex, and it is too context-specific to be interpreted by machines. As a result, scholars are often forced to limit their research to overly small data sets, or to study large data sets with ‘black box’ technology that might not be fit for purpose.
Our Information Access group designs transparent algorithms that explicitly include a ‘human in the loop’, so users and machines can interpret large quantities of data together. We make both the data and its processing open for critical inspection by scholars in the humanities and social sciences, and we develop metrics to measure its fitness for representative tasks, and visualize its limitations and biases for end users.
Research by the Information Access group brings the established tradition of historical source criticism into the digital age. Our Datalab platform takes the complexity and application context of the data into account, while our metrics help scholars to select the best tools for their research task and to assess the impact of technological bias on their research outcomes.
Contact person: Jacco van Ossenbruggen
Research group: Information Access (IA)
Research Partners: ADS, Rijksmuseum, National Library, Beeld & Geluid, Spinque, eHumanities.nl