Antiquities and the Machine: computational methods for the study of the representation of historical sites on Ordnance Survey Maps of Great Britain

Valeria Vitale, Katherine McDonough (Alan Turing Institute), Yao-Yi Chiang, Jina Kim, Zekun Li (University of Minnesota), Rainer Simon (Austrian Institute of Technology), Deborah Holmes-Wong (University of Southern California)

All welcome. Attend in-person in Senate House room 349, or watch live at https://youtu.be/Lcdb1xjvVFw

The Ordnance Survey (OS) is Great Britain’s national mapping agency. In the 18th and 19th centuries its maps have played an important role in shaping the nation. One unusual feature of these maps, that makes them especially interesting to classicists and archaeologists, is that, along with information about settlements, buildings or cultivated land, OS maps record the location of antiquities, both extant and disappeared, and even archaeological findspots. This practice implies that OS maps are committed to represent not only the visible features of the landscape, but even the invisible ones, blending different temporal layers in a single, complex document.

But how to deliver so much information in one image? OS maps combine different codes, making the most of polysemic signs. In particular, a separate system to represent antiquities and historical sites relies on the combination of ad hoc labels and special fonts. Three separate fonts are used to identify the time period of each historical place: pre-Roman (prehistoric to Anglo Saxon), Roman, and Norman (to Medieval).

The information about antiquities that can be found on OS maps is not only interesting from a semiotics perspective, for its elegant use of intersecting codes to deliver a large amount of information, but it is also a valuable resource for classicists and historians. These unusual labels on maps produced mostly during the Victorian period could be used to investigate the distribution of the antiquities, and what bias it shows. The selection of what was deemed worthy of appearing on the national map could be a useful key to understand, for example, how was Scottish heritage regarded, and how narratives about Roman presence in Britain were absorbed into British own imperialist attitudes.

Machines Reading Maps is an international collaboration between the Alan Turing Institute, the University of Minnesota, the Austrian Institute of Technology, and the University of Southern California. The project uses a combination of machine-learning and semantic technologies to detect and collect text on digitised maps, and to link it with external knowledge bases such as WikiData or OpenStreetMap. We want to apply this novel approach to the understanding of Victorian reception of antiquities, using the OS maps as a mirror of contemporary ideas and sensibilities. In the seminar we plan to discuss how to translate humanities research questions into machine-learning tasks that will enable us to process a large number of maps, creating new datasets and, ideally, delivering new answers.