The classification of space-time behaviour patterns in a British city from crowd-sourced data

Abstract

The use of social messaging as a means to represent activity and behaviour patterns across small geographical areas is explored. A large corpus of messages provides the source from which a range of interesting marker words are identified. Profiles of the variations in language across neighbourhoods can then be constructed. Areas are classified on the basis of the types of messages which they tend to generate. The resulting patterns are interpreted as suggesting that variations in behaviour and activity over time within an urban area are an important adjunct to well-established spatial variations. It is asserted that further elaboration of these promising investigations within appropriate analytic frameworks could extend our understanding of movement and behaviour patterns in cities in important ways. © 2013 Springer-Verlag Berlin Heidelberg.