A novel approach to mining travel sequences using collections of geotagged photos

In this paper we present a novel approach for analyzing the trajectories of moving objects and of people in particular. The mined data from these sequences can provide valuable information for understanding the surrounding locations, discovering attractive place or mining frequent sequences of visited places. Based on geotagged photos, our framework mines semantically annotated sequences. Our framework is capable of mining semantically annotated sequences of any length to discover patterns that are not necessarily immediate antecedents. The approach consists of four main steps. In the first step, every photo location is semantically annotated by assigning it to a known nearby point of interest. In the second step, a density-based clustering algorithm is applied to all unassigned photos, creating regions of unknown points of interest. In the third step, a travel sequence of every individual is built. In the final step, travel sequence patterns are mined using the semantics that were obtained from the first two steps. Case studies of Guimarães, Portugal (where the conference takes place) and Berlin, Germany demonstrate the capabilities of the proposed framework.