Crossing Silos: An Experiment in Reconciling to the LCNAF Using Additional Attributes
Common practice in name authority reconciliation with the Library of Congress’s Name Authority File (LCNAF) focuses on finding similarities between two names. Once a list of possible name matches is generated, most often in OpenRefine, significant manual review is necessary in order to confirm whether these pairings represent actual matches. In this presentation, we introduce our attempt to improve the reconciliation process by using additional fields from local faculty profiles and attributes (along with citations) from the LCNAF. We devised a series of tests in Python using datapoints for organizational affiliations with employer and educational institutions, a set of faculty-related occupations, and the person’s field of activity. In some cases, we also queried citations when newer fields were not present or if data was not found in them.
We will address complications we ran into when creating the project, share our assessment of which fields were the most useful in finding matches, and provide recommendations for replicating the work with your own faculty datasets. Finally, we will also offer insights from an evaluation performed on records which did not produce a match during these tests but which were later found through traditional name-name reconciliation processes. This presentation may also be of interest to those creating name authority records who wish to see real-world usage of this data.