Detecting and Identifying Collective Phenomena within Movement Data

Abstract

Collective phenomena are ubiquitous in our every day lives; each day we are likely to observe or take part in a collective. Examples include a traffic jam on the way to work, a flock of birds in the sky or a queue in the shop. These examples include only three types of collective that are considered in this thesis: those phenomena whose individual members can be assigned a physical location in geographic space. However, this criterion is satisfied by many different types of collective. The movement patterns that are exhibited by collectives are one of their most prominent properties; it is often the property that we wish to reason about most. For example, the movement patterns of crowds, traffic or demonstrations. This thesis hypothesises that, given a dataset that comprises the movement data for a group of individuals, the presence of certain collectives can be achieved through an examination of the exhibited movement patterns. To identify the different types of collective that exist, a general taxonomy of collectives is presented. A class of collectives are found to manifest themselves through spatial coherence. Therefore, a set of spatial coherence criteria have been developed that can be applied to a movement dataset to indicate if any individuals within that dataset may be participating in a spatial collective. To indicate the different types of spatial collective that may be extracted, a taxonomy of spatial collectives is also presented.