School of
Information Technology and Electrical Engineering

Quality-aware Trajectory Processing Using Significant Locations

Speaker: 
Han SuThu, 29/08/2013 - 10:30
Venue: 
49-502
Host: 
Prof Xiaofang Zhou
Abstract: 

Due to the prevalence of GPS-enabled devices, wireless communications and sensing technologies, spatial trajectories that describe the movement histories of moving objects are being generated and accumulated with an unprecedented pace. Querying and mining these large scale trajectories are the basis of many location-based services, such as route recommendation. However, given the fact that a trajectory is a discrete approximation of the original continuous path by sampling the locations periodically, the trajectories in a practical database could be derived by quite different sample strategies, resulting in a set of heterogeneous trajectory data. The heterogeneity of trajectory data has negative impact on the effectiveness on k-nearest-neighbour (KNN) search, which is the most widely used trajectory query.

During the PhD study, a systematic approach to transform a heterogeneous trajectory dataset to the one with (almost) unified sample strategies, which is called trajectory calibration, has been proposed. Specifically, we propose the method to rewrite trajectories using landmarks, which are fix objects in space, instead of sample points. Using the landmark to describe trajectory, the quality of trajectory queries can be improved a lot: popular trajectory query can be benefited by analysing landmarks on trajectories; the efficiency of most trajectory queries can be improved dramatically since the landmark-based trajectory can be indexed in totally different way comparing with previous trajectory index structures, such as R-tree. What's more, by analysing a specific person's moving history and social information, we can provide personalization landmarks for her, which makes her trajectory query results more accurate.

Biography: 

Han Su is currently a PhD student with the Data Engineering and Pattern Recognition Research Division at the University of Queensland. She obtained her Bachelor in Software Engineering degree from Nanjing University, China, in 2011. Her research interests include efficient spatial trajectory query processing and quality enhancing. 

Seminar Type: 

PhD Confirmation Seminar

Pages