Semantic road networks for recommender systems
Thomas A. Neirynck · 2012
In visualizations of non-spatial data, the distance similarity metaphor can rarely be preserved at all scales, inhibiting correct judgment of similarity and connectivity between features on a semantic map. The author shows that roads in geographic space are useful indicators of landscape topography, of connectivity and of importance of features across all scales. The author proposes a novel method to generate road-like networks on a Self Organizing Map and illustrates the use of such Semantic Road Networks in the context of a recommender system. The author argues that interactive maps of these Semantic Road Networks improve transparency of the system, and enable a new method of generating recommendations by traveling along routes on a semantic map.