After Prosopography? Data modelling, models of history, and new directions for a scholarly genre
Timothy Hill (King’s College London)
Digital Classicist and Institute of Classical Studies Seminar 2010
Friday June 18th at 16:30, in room STB9, Senate House, Malet Street, London WC1E 7HU
This paper explores the potential of new data technologies to deal tractably with two characteristics that have marked the scholarly genre of the prosopography throughout its evolution–and, in the process, to extend the capacities of the genre into new areas of historical and scholarly enquiry.
The first of these characteristics is a drive towards comprehensiveness–not only in the sense that the scale of prosopographies has grown as the genre has matured, but also in that the kinds of entities this information concerns have tended to diversify; whereas, e.g., the Prosopographia Imperii Romani (PIR) (1897/1936), consists straightforwardly of a list of Roman office-holders, more recent works have embraced everything from cuneiform tablets to breweries.
The second persistent characteristic of the genre has been a division between the data model implicitly employed in compiling prosopographies, and the research interests of the historians who use them. Typically researchers have turned to prosopographies in order to answer questions about institutions and social structures; while what prosopographies in fact tend to provide is information about individuals. In data-modelling terms, historians might be said to be interested in relational information, while prosopographers are largely focused upon attribute data.
Until recently, technological limitations have tended conservatively to reinforce an anthropocentric and attribute-data orientation in prosopography–the cumbrousness of paper media and formalism of relational database technologies serving to limit the range of entities modelled and the number of high-level relationships asserted of them.
Graph-based storage formats such as RDF, however, are now sufficiently viable to provide a unifying data model for prosopographers; and while the basis for this unification–their underlying graph model–is technical and abstract, its immediate implications are potentially dramatic.
The most fundamental of these implications is that, because in a graph-based data model all entities, including attributes and their values, are treated as nodes in a network, there is no longer any reason to privilege any entity within the prosopography as central. Within a connected network one can begin exploration at any point, and a graph model thus comfortably accommodates the prosopographical tendency to incorporate an ever-expanding range of entities.
Furthermore, this subsumption of attribute data into a graph model makes RDF and similar data formats amenable to techniques of network analysis of considerable interest to the historian. Network techniques are well-suited to the sparse, irregular data-sets characteristic of ancient sources – and, when taken in combination with ontological tools, open up the possibility of new forms of historical enquiry, including but not limited to: the automated deduction and detection of social cliques and structures; the experimental testing of causal models against known events; and formalised comparative analysis of cultural phenomena across periods, localities, and societies. Such techniques have begun recently to inform new research approaches in fields as diverse as systems ecology and anthropology, and there is reason to believe that they may prove even more fertile for historians. For while conventional network analysis is typically acausal, atemporal and relationally monodimensional, such restrictions need not necessarily apply to historically-based networks.
The seminar will be followed by wine and refreshments.