Finding All Co-occurrences from Spatiotemporal Data in an Efficient and Scalable Manner

Abstract

People are constantly moving around, coming and going from one place to another. People interact with friends, coworkers, and strangers on a daily basis. This data - that is a person’s location at a given time - has the potential to tell much about the person’s social connections. This has broad-ranging applications: it has applications in fields such as epidemiology, criminology, and marketing. Previous work on computing co-occurrences from spatiotemporal data to infer social ties has been mostly mathematical and has not focused on achieving computationally efficient results. In addition, many of the approaches have relinquished finding every co-occurrence in order to compute things in an efficient and scalable manner. This work sets out to discuss a scalable and computationally efficient approach to compute all co-occurrences from a spatiotemporal dataset for the purpose of inferring social ties.