Towards a Semantic Trajectory Similarity Measuring

Objectives: To propose a new similarity function to determine trajectory similarity considering semantic aspects. Methods/Analysis: We propose different methods to calculate the similarity according to visited sites or activities performed: the first one considers only the sites included in the trajectories and the second considers the activities performed by the trajectories in the sites. A third method is proposed to find the similarity bitten trajectories based on both sites and activities. Findings: The similarity measure presented in this work allows us to make comparisons and user analysis according to trajectory data generated by users, which represents their routines, likes and preferences. This could be a key element for recommender systems, clustering or social networks. Novelty/Improvements: Our methods consider semantic aspects for finding the similarity of trajectories, considering visited sites and activities performed in these sites.