Go Beyond Raw Trajectory Data: Quality and Semantics

Past decades have witnessed extensive studies from both academia and industries over trajectory data, which are generated from a diverse range of applications. Existing literature mainly focuses on raw trajectories with spatio-temporal features such as location, time, speed, direction and so on. Recently, the pervasive use of smart mobile devices like smart phones, watches and bands have brought about more generation of trajectory by personal users (instead of companies or organizations) and from online space (instead of physical space), where individuals can decide when and where to log on and share their locations with others. The more discentralized and contextualized trajectory sources have brought some unique challenges for database management with respect to the quality and semantics of trajectories data. With more applications and services relying on trajectory data analysis, it is necessary for us to think about how these new issues will affect the traditional way that trajectories are digested and processed. In this paper we will elaborate on these challenges and introduce our recent progress in the respective directions. The message we try to deliver is that raw trajectories themselves no longer satisfy the requirement of today’s mainstream applications. To really release the power of trajectory-based applications, we should go beyond the raw trajectory data by enhancing their quality and semantics, which calls for novel computing architectures, paradigms and algorithms with sufficient capabilities to manage and analyse the enhanced trajectory data.