The analysis of sensible data requires a proper anonymization of values in order to preserve the privacy of individuals. Information loss should be minimized during the masking process in order to enable a proper exploitation of data. Even though several masking methods have been designed for numerical data, very few of them deal with categorical (textual) information. In this case, the quality of the anonymized dataset is closely related to the preservation of semantics, a dimension which is commonly neglected of shallowly considered in related words. In this paper, a new masking method for unbounded categorical attributes is proposed. It relies on the knowledge modeled in ontologies in order to semantically interpret the input data and perform data transformations aiming to minimize the loss of semantic content. On the contrary to exhaustive methods based on simple hierarchical structures, our approach relies on a set of heuristics in order to guide and optimize the masking process, ensuring its scalability when dealing with big and heterogenous datasets and wide ontologies. The evaluation performed over real textual data suggests that our method is able to produce anonymized datasets which significantly preserve data semantics in comparison to apporaches based on data distribution metrics.
Weitere Kapitel dieses Buchs durch Wischen aufrufen
Bitte loggen Sie sich ein, um Zugang zu diesem Inhalt zu erhalten
Sie möchten Zugang zu diesem Inhalt erhalten? Dann informieren Sie sich jetzt über unsere Produkte:
- Ontology-Based Anonymization of Categorical Values
- Springer Berlin Heidelberg
ec4u, Neuer Inhalt/© ITandMEDIA