BiographyNet: historical queries, text analysis and visualization
The Biography Portal of the Netherlands hosts over 125,000 biographies on over 76,000 individuals from the present day Netherlands. A wide variety of biographical dictionaries, from the eighteenth to the twenty-first century, containing these biographies is included in the Portal. The Portal therefore hosts a rich potential for historical research. At the moment however, it is not much more than a handy reference work for historians and other humanities scholars. The BiographyNet team at the VU, in collaboration with Huygens ING and the eScience Center, aims at creating a tool to intelligently query the data for historical research.
There are several challenges which have to be faced when creating such a tool. What kind of historical questions can or cannot be (fully) answered with the help of computational methods? How do you create a complex tool that still provides the historian insight into how it works? How can you break up complex historical questions into software interpretable queries? What kind of precision and recall can we achieve in unstructured text from different centuries? How do we deal with archiving queries and provenance of the data and manipulation of the data? What kind of interface can we create that inspires historians without limiting them to a few basic research questions? The lecture will present these issues for a discussion at the intersection of historical theory, computational linguistics and computer science.