Multidimensional Integration of RDF Datasets

Abstract

Data providers have been uploading RDF datasets on the web to aid researchers and analysts in finding insights. These datasets, made available by different data providers, contain common characteristics that enable their integration. However, since each provider has their own data dictionary, identifying common concepts is not trivial and we require costly and complex entity resolution and transformation rules to perform such integration. In this paper, we propose a novel method, that given a set of independent RDF datasets, provides a multidimensional interpretation of these datasets and integrates them based on a common multidimensional space (if any) identified. To do so, our method first identifies potential dimensional and factual data on the input datasets and performs entity resolution to merge common dimensional and factual concepts. As a result, we generate a common multidimensional space and identify each input dataset as a cuboid of the resulting lattice. With such output, we are able to exploit open data with OLAP operators in a richer fashion than dealing with them separately.