Certainty and Uncertainty in Representing Family Histories

December 9, 2014

Luther Tychonievich will present his data model for collaborative genealogy (and its attendant unreliable data), this Friday, December 12, at 2pm in Alderman 421.

Abstract: Most family history data is stored as a set of "facts" with human-targeted textual citations and notes. All the computer understands in these models is a single internally-consistent view of the past. However, anyone who has done research in any field knows that evidence is almost always messy and ambiguous, potentially supporting many conclusions. Failing to store this ambiguity can lead to edit wars, ignoring valid leads, and following dead ends much longer than needed.

This presentation will provide an overview of polygenea, a data model designed to represent what we find in research in a way that admits differing opinions, mutually contradictory possibilities, and indirect and negative evidence. We'll explore how this model streamlines collaboration, simplifies researcher privacy, supports explorations of many alternatives simultaneously, and removes language dependence from much of the data. Time permitting, we may also discuss its application in other disciplines, the design of wrappers to extract one or more traditional "fact" data sets from the underlying data, and support for vendor-controlled standardization and automated format correction.

Luther Tychonievich is a lecturer in the UVA CS department. His research interests include facilitating large-scale cooperative family history research.