Parsing and Matching Dates in VIAF

Abstract

The Virtual International Authority File (OCLC Online Computer Library Center 2013) http://viaf.org is built from dozens of authority files with tens of millions of names in more than 150 million authority and bibliographic records expressed in multiple languages, scripts and formats. One of the main tasks in VIAF is to bring together personal names which may have various dates associated with them, such as birth, death or when they were active. These dates can be quite complicated with ranges, approximations, BCE dates, different scripts, and even different calendars. Analysis of the nearly 400,000 unique date strings in VIAF led us to a parsing technique that relies on only a few basic patterns for them. Our goal is to correctly interpret at least 99% of all the dates we find in each of VIAF’s authority files and to use the dates to facilitate matches between authority records. Python source code for the process described here is available at https://github.com/OCLC-Developer-Network/viaf-dates.