Abstract
This paper presents a framework for intrinsic point of interest discoveryfrom trajectory databases. Intrinsic points of interest are regions of ageospatial area innately defined by the spatial and temporal aspects oftrajectory data, and can be of varying size, shape, and resolution. Anytrajectory database exhibits such points of interest, and hence are intrinsic,as compared to most other point of interest definitions which are said to beextrinsic, as they require trajectory metadata, external knowledge about theregion the trajectories are observed, or other application-specificinformation. Spatial and temporal aspects are qualities of any trajectorydatabase, making the framework applicable to data from any domain and of anyresolution. The framework is developed under recent developments on theconsistency of nonparametric hierarchical density estimators and enables thepossibility of formal statistical inference and evaluation over such intrinsicpoints of interest. Comparisons of the POIs uncovered by the framework insynthetic truth data to thousands of parameter settings for common POIdiscovery methods show a marked improvement in fidelity without the need totune any parameters by hand.