3.5. Conclusion

Efficiently discovering candidate datasets for linking and the links between these candidates are both challenging tasks, given the size and diversity of the Web of data. In this chapter, we first focus on the identification of candidate datasets for data linking where we provide an overview of various existing approaches leading to a comprehensive discussion on the topic. Given a source data set to be linked, and once the target dataset is identified, it is important to deal with different heterogeneity problems that may occur between these two datasets, such as differences in descriptions on the value, ontological or logical level in order to compare the resources they contain efficiently. In this context, we identified and then provided the possible solutions to these heterogeneities that exist in the literature.

The datasets connected by owl: sameAs links form a non-oriented graph of very large size. The absence of strong connectivity (due to lack of owl: sameAs links...