Data Warehouses and the Semantic Web - Data Warehouse Systems: Design and Implementation - page 528

Chapter 14
Data Warehouses and the Semantic Web
The availability of enormous amounts of data from many different domains
is producing a shift in the way data warehousing practices are being carried
out. Massive-scale data sources are becoming common, posing new challenges
to data warehouse practitioners and researchers. The semantic web, where
large amounts of data are being stored daily, is a promising scenario for data
analysis in a near future. As large repositories of semantically annotated data
become available, new opportunities for enhancing current decision-support
systems will appear. In this scenario, two approaches are clearly identified.
One focuses on automating multidimensional design, using semantic web
artifacts, for example, existing ontologies. In this approach, data warehouses
are (semi)automatically designed using available metadata and then popu-
lated with semantic web data. The other approach aims at analyzing large
amounts of semantic web data using OLAP tools. In this chapter, we tackle
the latter approach, which requires the definition of a precise vocabulary
allowing to represent OLAP data on the semantic web. Over this vocabulary,
multidimensional models and OLAP operations for the semantic web can
be defined. Currently, there are two proposals in this direction. On the one
hand, the data cube vocabulary (also denoted QB) follows statistical data
models. On the other hand, the QB4OLAP vocabulary follows closely the
classic multidimensional models for OLAP studied in this topic.
In this chapter, we first introduce in Sect. 14.1 the basic semantic web
concepts, including the RDF and RDFS data models, together with a study of
RDF representation of relational data and a review of R2RML, the standard
language to define mappings from relational to RDF data. In Sect. 14.2 ,we
give an introduction to SPARQL, the standard query language for RDF data.
In Sect. 14.3 , we discuss the representation and querying of multidimensional
data in RDF, including an in-depth discussion of the QB and QB4OLAP
vocabularies. We continue in Sect. 14.4 showing how the Northwind data
cube can be represented using both vocabularies. We conclude in Sect. 14.5
by showing how to query the QB4OLAP representation of the Northwind
data warehouse in SPARQL.