Similarity search over enriched geospatial data

Abstract

Enriched geospatial data refers to geospatial entities associated with additional information from various sources, such as textual, numerical or temporal. Exploring such data involves multi-criteria search and ranking across several heterogeneous attributes. In this paper, we model this task as a rank aggregation problem. Our method automatically scales similarity scores across diverse attributes without relying on user-specified parameters. It also allows to retrieve and combine information from multiple sources during query execution. We evaluate our approach using a large real-world dataset of enriched geospatial entities representing news articles.