Pointers Extraction of Trajectory Data for Semantic Knowledge Discovery

People like services that can help them undertake their daily activities more efficiently. Mobile technologies have enabled deployment of a variety of Internet–based services within the realm of location-based services (LBS) (Steiniger, Neun, and Edwardes 2006). The adoption of these technologies has led to mammoth amounts of trajectory data. To use these services effectively, analysis of LBS data across different application domains is required to identify similar behaviour or discover regularities between users that can be used to predict user's future behaviour (Nanni et al. 2008). Several algorithms have been developed that mine trajectory data specifically for the discovery of periodic movement (Cao, Mamoulis, and Cheung 2007), trajectory classification (Lee et al. 2008), and relative motion patterns (Laube, Imfeld, and Weibel 2005). These algorithms have focused on the geometric properties of trajectories (Bogorny et al. 2011), which are good at discovering patterns, but do not always aid behavioural interpretation. According to Dodge, Weibel & Lautenschütz (2008) movement behavior depends on the context of the movement; where movement happens, what time of day, what day of the week, etc. Thus, methods for interpreting the semantics of context within the knowledge discovery process can lead to discovery of semantic trajectory patterns (Ong et al. 2010). A few researchers such as Borgorny et al. (2009), Baglioni et al. (2009) and Trasarti et al. (2010) have considered semantic information as a means of helping understand trajectory patterns. However, little attention has been given to the preprocessing phase to extract activities and other relevant information that can aid semantic interpretation. This paper proposes a methodology for interpreting moving object behavior by considering types of pointers: stop frequency; and stop duration, to infer moving object activities. Then, an association rule algorithm is used to extract meaningful, understandable, and useful patterns from trajectories.