3.3. Challenges of linking data

In previous sections, we have seen different methods which are proposed for the selection of candidate datasets for linking with a given (presumably yet unpublished) RDF graph. Provided that this task has been performed successfully, we now have at hand two RDF graphs to link, or else to discover the related entities across these two graphs. A relation of particular interest for the Semantic Web community is that of identity, given by the OWL predicate owl: sameAs. However, a given real-world entity may be described differently or even with complementary information in different data sources. Hence, the automatic discovery of identity links may become particularly difficult considering the heterogeneity of data on the Web. In this section, we focus on a number of real-world issues that may arise when comparing resources across datasets, and we illustrate these issues by the means of a fictional example given in Figure 3.3.

Let us imagine that the composer...