The Effect of Space-Language Bias on Toponym Co-occurence Derived Networks

Abstract

Toponym co-occurrence analysis on unstructured sources has been suggested as a possible method for obtaining data for the study of urban networks. This method is particularly beneficial for mod- elling international relations and the relations between smaller places. However, it also suffers from potentially introducing space-language bias into the networks. This paper creates a network of 151 European cities derived from English and French versions of Wikipedia using toponym co-occurrence analysis to be used as a case study. City-pairs in the English and French language sphere are found to have a tendency to be over-represented in the respective data sources compared to the patterns expected from gravity modelling. Nevertheless both of the resulting networks fit expected patterns, showing the applicability of toponym co-occurrence analysis.