Search for: All records

Geocoding is a fundamental yet complex step in temporal studies due to constantly evolving administrative borders and place names, and the uncertainty of geographic and temporal information. For example, identifying locations within crowdsourced datasets, such as family trees, is complex because recorded place names may be uncertain, inaccurate, and contain varying spellings (for instance, full names or abbreviations) and in inconsistent formats, such as mentioning only the country, or state, or a combination of city, county, state, and country. Moreover, place names and administrative boundaries drastically change over time, adding another layer of complexity to the geocoding process of fine-scale places. This paper presents a workflow for geocoding birthplaces of US-born individuals from crowd-sourced genealogical files spanning from 1789 to 1940. We introduce a method that geocodes the birth locations at the finest possible level by matching places with corresponding historical administrative boundaries within a range of individuals’ birth years. Our preliminary study analyzing 72,335 trees with over 250 million individual records shows the potential of our approach for use in complex crowd-generated spatio-temporal datasets.