An ETL Framework for Online Analytical Processing of Linked Open Data

Abstract

AbstractGrowing amount of data are being published online in machinereadable formats, and LOD (Linked Open Data) has emerged as a way to share such data across Web resources. Since LOD data often contain numerical data, such as statistics, there is a growing demand to make OLAP (Online Analytical Processing) analysis over such data. To make it possible to apply off-the-shelf OLAP systems for analyzing LOD data, we propose a framework to streamline the Extract, Transform, and Load (ETL) process from LOD to multidimensional data models for OLAP. Unlike other related approaches, our framework does not require RDF vocabularies dedicated for specifying multidimensional model for OLAP. Instead, given an LOD dataset, we exploit the relationships among entities and external information in the referenced LOD to generate an OLAP data model. In a case study, we demonstrate that our framework can extract OLAP data models from different kinds of real LOD datasets.KeywordsLinked Open DataOLAPETLMultidimensional Model