Motivated by the proliferation of trajectory data produced by advanced GPS-enabled devices, trajectory is gaining in complexity and beginning to embroil additional attributes beyond simply the coordinates. As a consequence, this creates the potential to define the similarity between two attribute-aware trajectories. However, most existing trajectory similarity approaches focus only on location based proximities and fail to capture the semantic similarities encompassed by these additional asymmetric attributes (aspects) of trajectories. In this paper, we propose multi-aspect embedding for attribute-aware trajectories (MAEAT), a representation learning approach for trajectories that simultaneously models the similarities according to their multiple aspects. MAEAT is built upon a sentence embedding algorithm and directly learns whole trajectory embedding via predicting the context aspect tokens when given a trajectory. Two kinds of token generation methods are proposed to extract multiple aspects from the raw trajectories, and a regularization is devised to control the importance among aspects. Extensive experiments on the benchmark and real-world datasets show the effectiveness and efficiency of the proposed MAEAT compared to the state-of-the-art and baseline methods. The results of MAEAT can well support representative downstream trajectory mining and management tasks, and the algorithm outperforms other compared methods in execution time by at least two orders of magnitude.

Map matching is a technique used to identify which path a vehicle is travelling on in a road map. Since it is a crucial fundamental step for a wide range of transportation applications, many map-matching algorithms have been devised ranging from simple geometric calculation methods to more sophisticated methods. However, the study of spatial relationship patterns between GPS coordinates and road segments mapped with map-matching has not received enough attention from researchers. This paper presents a framework, called Proxy Map Matching (PMM), to learn such patterns using machine learning techniques. However, we find that solely employing machine learning techniques on such data is not sufficient to capture the patterns. Solving this problem that way results in an inaccurate proxy model. In PMM, we construct several proxy map matchers and assign them to each group of data based on their spatial proximity, thereby achieving accuracy improvement. An experiment on real-world data shows that the framework achieves above 85% accuracy with integration of machine learning techniques, and outperforms the methods which solely employ machine learning techniques significantly. Moreover, the proposed proxy model can perform very fast matching. For 14,177 GPS coordinate pairs per second, we achieve 88.4% accuracy.