|
Abstract:
|
Much of the work in statistical crime prediction has used static geographical information, perhaps in combination with time, to forecast where and when crime is most likely to occur. This work has not investigated micro-level movement patterns of individuals in the area of interest. Although it is difficult to collect fine-grained movement data directly from individuals' devices, it is not so difficult to obtain massive collections of textual data generated by users of social media services such as Twitter. When geotagged, this content implicitly describes movement patterns for many individuals. This talk will present recent research on using these patterns to augment traditional statistical crime prediction methods. The talk will focus on (1) the use of spatiotemporally tagged Twitter posts for inferring micro-level movement patterns, and (2) the use of these patterns to predict the spatiotemporal occurrence of actual crime from a large U.S. city.
|
