Querying the Digital Archive of Science: Distant Reading, Semantic Modelling and Representation of Knowledge

The archive of science is a place where scientific practices are sedimented in the form of drafts, protocols of rejected hypotheses and failed experiments, obsolete instruments, outdated visualizations and other residues. Today, just as science goes more and more digital, so does its archive, giving rise to new research practices and opening new frontiers of knowledge for the historian (from big data to the longue durée). These collections clearly differ from the traditional lieux de mémoire. What they store are not tangible and authentic objects, but data to be processed and interpreted by computer algorithms and software. The way archival data is situated, described and presented to the user is prefigured and mediated by digital technologies and infrastructures. How do these new digital infrastructures operate and shape our encounter with the scientific past? What can we learn about the science of the past from its residues as they go digital and turn into data? And how could these collections be made meaningful for the queries of both historians and the wider public?I argue that the digital archive does more than store some remnants of the past; it becomes an active agent in their interpretation. For this reason, we need to explore the limits, conditions, and affordances of the interpretations it offers and makes possible. This dissertation probes into how we understand and interpret the past of science through its digital archive, focusing on its specific modes of representation, the methods of treating the past it offers, and its transmission mechanisms. Based on a large corpus of scientific collections and mixing quantitative and qualitative approaches, the study assembles the elements of a humanist (instead of engineering-oriented) ontology for the scientific archive, transferring concepts and perspectives from the history of science into computational language. Experimenting with the methods offered by the digital (distant reading, semantic modelling) and the interpretations they enable, this dissertation reimagines the digital archive as a way of making the past (of science).