Clustering Pipelines of Large RDF POI Data

Abstract

Among the various domains using large RDF graphs, applications often rely on geographical information which is often represented via Points Of Interests. In particular, one challenge is to extract patterns from POI sets to discover Areas Of Interest (AOIs). To tackle this challenge, a typical method is to aggregate various points according to specific distances (e.g. geographical) via clustering algorithms. In this study, we present a flexible architecture to design pipelines able to aggregate POIs from contextual to geographical dimensions in a single run. This solution allows any kind of clustering algorithm combinations to compute AOIs and is built on top of a Semantic Web stack which allows multiple-source querying and filtering through SPARQL.