A Survey on Spatiotemporal and Semantic Data Mining

The wide proliferation of GPS-enabled mobile devices and the rapid development of sensing technology have nurtured explosive growth of semantics- enriched spatiotemporal (SeST) data. Compared to traditional spatiotemporal data like GPS traces and RFID data, SeST data is multidimensional in nature as each SeST object involves location, time, and text. On one hand, mining spatiotemporal knowledge from SeST data brings new opportunities to improving applications like location recommendation, event detection, and urban planning. On the other hand, SeST data also introduces new challenges that have led to the developments of various techniques tailored for mining SeST information. In this survey, we summarize state-of-the-art studies on knowledge discovery from SeST data. Specifically, we first identify the key challenges and data representations for mining SeST data. Then we introduce major mining tasks and how SeST information is leveraged in existing studies. Finally, we provide an overall picture of this research area and an outlook on several future directions of it. We anticipate this survey to provide readers with an overall picture of the state-of-the-art research in this area and to help them generate high-quality work.