Mining Sub-trajectory Cliques to Find Frequent Routes (Technical Report) Htoo Htet Aung, Long Guo and Kian-Lee Tan School of Computing, National University of Singapore Abstract. Knowledge of the routes frequently used by the tracked ob- jects are embedded in the massive trajectory databases archiving spatial- temporal movement data of the objects in question. Such knowledge has various applications in optimizing ports’ operations, understanding wild-life behaviours, and navigation/route-recommendation systems but is difficult to extract in many real-life scenarios, where the underlying road network information is not available. We propose a novel approach, which discovers frequent routes without any prior knowledge of the un- derlying road network by mining sub-trajectory cliques. Since mining all sub-trajectory cliques is an NP-Complete problem, we proposed two ap- proximate algorithms based on the Apriori algorithm. Empirical results showed that our algorithms can run faster than the existing approxima- tion algorithm appeared in [1] and provide a tighter results. Our exper- iments also showed that the frequent routes reported by our algorithms are intuitive. 1 Introduction Advances in location-tracking technologies, such as the Global Positioning System (GPS), enable access to spatial-temporal movement data of the tracked objects in question. Such movement data are usually archived in Trajectory Databases (TJDBs) for further analysis to discover actionable knowledge and support deci- sion making. For instance, the Automatic Identification System (AIS) transmits the spatial-temporal movement data of a ship — or the ship trajectory — to maritime authorities, who use it to track and monitor the movement of the ves- sels in their territories [2]. The authorities often archive the ship trajectories for further studies to obtain actionable knowledge, which is, in turn, used to opti- mize their ports’ operations. Similarly, businesses in the public transportation industry (taxi and bus operators) and those in the logistics industry record and archive the movement data of their fleets in TJDBs for analysis aiming to im- prove the quality of their services. In addition, ecologists and marine biologists are looking forward to track and record the trajectories of the animals they are studying [3]. Since spatial-temporal movement data (trajectories) of the tracked objects are archived in TJDBs, the tracks taken by the entities in question are also