RiAiR: A Framework for Sensitive RDFProtection

Abstract

The Semantic Web and the Linked Open Data (LOD) initiatives promote the integration and combination of RDF data on the Web. In some cases, data need to be analyzed and protected before publication in order to avoid the disclosure of sensitive information. However, existing RDF techniques do not ensure that sensitive information cannot be discovered since all RDF resources are linked in the Semantic Web and the combination of different datasets could produce or disclose unexpected sensitive information. In this context, we propose a framework, called RiAiR, which reduces the complexity of the RDF structure in order to decrease the interaction of the expert user for the classification of RDF data into identifiers, quasi-identifiers, etc. An intersection process suggests disclosure sources that can compromise the data. Moreover, by a generalization method, we decrease the connections among resources to comply with the main objectives of integration and combination of the Semantic Web. Results show a viability and high performance for a scenario where heterogeneous and linked datasets are present.