Searching for data sources for the semantic enrichment of trajectories

The fast growing number of datasets available on the Web inspired researchers to propose innovative techniques to combine spatiotemporal data with contextual data. However, as the number of datasets has increased relatively fast, finding the most appropriate datasets for enrichment also became extremely difficult. This paper proposes an innovative approach to rank a set of datasets according to the likelihood that they contain relevant enrichments. The approach is based on the intuition that the sequence of places visited during a trajectory can induce the best datasets to enrich the trajectory. It relies on a supervised approach to learn rules of association between visited places and meaningful datasets.