“…Recently, there has been a surge in semantic enrichment (Zheng, 2015) with GPS trajectory data. Many studies have described semantic information from moving object data, including stop‐and‐move models (Damiani & Güting, 2014; Parent, Pelekis, Theodoridis, & Yan, 2013; Spaccapietra & Parent, 2011), time‐dependent‐label models (Damiani, Issa, Guting, & Valdes, 2014; Güting, Valdés, & Damiani, 2015; Renso, Baglioni, de Macedo, Trasarti, & Wachowicz, 2013), location‐correlation (Cai, Lee, & Lee, 2018; Valdés, Damiani, & Güting, 2013; Yan, 2011), and other hybrid methods (Issa, 2016; Wan, Zhou and Pei, 2017). Among these methods, raw trajectories are divided into sequences of subtrajectories that meet prescribed spatiotemporal thresholds, and semantic annotations are generated by a wide range of methods, such as clustering (Cai, 2017; Hung, Peng, & Lee, 2015; Yuan, Sun, Zhao, & Li, Wang, 2017), manual annotations (Cai, 2017; Nabo, Fileto, Nanni, & Renso, 2014), and spatial correlation methods (Renso et al, 2013; Wang & Kwan, 2018).…”