Combining link keys and similarity-based approaches to data interlinking

Data interlinking plays a crucial role in expanding and enriching linked open data. One approach to address data interlinking involves the utilization of link keys, which extends the concept of keys to scenarios where two RDF datasets are described using distinct ontologies. An alternative approach employs various similarities measures to evaluate the proximity of instances and decide if they have to be linked. In this work, we will employ numerical link specifications and document similarity. Both approaches have their respective advantages and limitations, and none of them provide a comprehensive solution for all interlinking challenges. Hence, this master thesis considers ways to combine such approaches elegantly, using the combination operators disjunction and injective complementation. We will make use of pairs of datasets to perform experiments in order to analyze the performance of different combination methods. The results show that the combination of the two approaches was productive, although it is very depending on the task we are analyzing.