Improving Link Specifications using Context-Aware Information

There is an increasing interest in publishing data using the Linked Open Data philosophy. To link the RDF datasets, a link discovery task is performed to generate owl:sameAs links. There are two ways to perform this task: by means of a classier or a link specication; we focus in the latter approach. Current link specication techniques only use the data properties of the instances that they are linking, and they do not take the context information into account. In this paper, we present a proposal that aims to generate context-aware link specications to improve the regular link specications, increasing the eectiveness of the results in several real-world scenarios where the context is crucial. Our context-aware link specications are independent from similarity functions, transformations or aggregations. We have evaluated our proposal using two real-world scenarios in which we improve precision and recall with respect to regular link specications in 23% and 58%, respectively.