Presenting on behalf of Department of Near Eastern Studies, University of California, Berkeley, by Adam G. Anderson and Laurie Pearce, on August 27, 17.00-17.20 in Session 2: Prosopographical work
ABSTRACT mārē BPS: the offspring of a prosopography project
Assyriologists and members of the Research IT team at the University of California, Berkeley teamed to build Berkeley Prosopography Services (BPS), an open-source, corpus agnostic toolkit to support the disambiguation of namesakes in text corpora. At the heart of the toolkit, the disambiguation engine accepted corpus-specific, user-defined rules that identified persons on the basis of naming conventions, compatible and incompatible roles, chronological features (including length of career) to compute the likelihood that two or more name instances referred to the same person. Once users collapsed disambiguated persons, the graph engine generated a real-time interactive social network visualization. Although BPS is now dormant (2009-2018), many of the foundational and innovative concepts have been adopted by subsequent prosopography projects under development in the Department of Near Eastern Studies and the Digital Humanities program at the University of California, Berkeley.
Niek Veldhuis and Adam Anderson lead “The Sumerian Networks DS-Discovery Project,” which demonstrates how a careful combination of historical sources with quantitative analysis can generate reproducible prosopographical models for a century of data from the Ur III archives (2100-2004 BCE). Working with the online databases of Ur III texts (BDTNS, ORACC & CDLI), the resulting network graphs juxtapose the roles and titles of each named entity within their social, institutional, geographical, and temporal substrates, and visualizes these entities in a relational network graph. Such networks of micro-economic transactions may then be measured statistically, leveraging mathematical formalization and geo-spatialization for macro-economic models. These results provide a map of the social-spatial distances of the individuals and locations of otherwise unknown names in the historical annals of the Ur III texts. The combination of these elements has enabled the progress toward unsupervised disambiguation of the Ur III prosopography, which in turn reveals latent organizational structures, resulting from the sum of individual actors, institutions, and geographic names, when modeled both spatially and temporally. The repository for the ongoing research is hosted on GitHub with Notebooks and Network Graphs.
The present paper demonstrates how The Sumerian Networks DS-Discovery Project is, in many ways, an intellectual “son of BPS”. Its unsupervised disambiguation of the Ur III prosopography implements features of the BPS disambiguation engine, built on the concept of the NRAD ( a Name in a Role in an Activity in a Document) as the primary criterion for disambiguating namesakes. The joint presentation of Adam Anderson and Laurie Pearce underscores the value of collaboration in extending the scope of research and enriching the development of tools in digital Assyriology.