Authors: Junjun Yin*, Social Science Research Institute, Institute for CyberScience, Penn State University
Topics: Spatial Analysis & Modeling, Geographic Information Science and Systems
Keywords: mobility, preferential return, semantic trajectory
Session Type: Paper
Start / End Time: 3:20 PM / 5:00 PM
Room: Bayside B, Sheraton, 4th Floor
Presentation File: No File Uploaded
Human mobility is a complex human behavior, which is confounded by human variability, by its nature difficulty to study. At the same time, understanding human mobility patterns has profound real-world applications ranging from urban planning, traffic management, and even studying the spread of infectious diseases. With large volume of movement datasets extracted from mobile phone records, GPS recorded trajectories, and even geo-located social media messages, it is possible to study mobility patterns of individuals with high spatial resolution and temporal granularity. However, in most of the current models and methods developed for seeking mobility patterns, user locations are simply treated equally as pairs of geographical coordinates. The lack of geographic context of user locations limits our ability to differentiate user movements regarding varying semantics, where the information regarding the place of the origin and destination of a user’s movement is absent. In this study, we utilize a data synthesis approach to enhancing the geographic context of geo-located Twitter user locations and enabling situation-awareness of the user movements. We augment each recorded Twitter user location with land use information (as semantics) by integrating detailed parcel-level land use maps of three US cities: Boston, Chicago and San Diego. Based on the enriched movement datasets, we studied the decomposition of the Twitter user locations and identified the preferential return behaviors at the parcel-level, both spatial and temporally. As we were able to annotate each movement in the dataset, we further capture the sequential mobility patterns that contribute to the preferential return behaviors.