Privacy enforcing framework in high dimensional data streams for real-time data markets

The thesis works on a new method called z-anonymity that aims to anonymize a continuous streaming of large amounts of data making decisions without delays based only on the last window of time. The whole work is an effort to achieve the best perfomance in terms of privacy level, computational time and usefulness of the anonymized data. The proposed algorithm combines several techniques used in the field, particularly the generalization of numeric and non-numeric attributes as a substitute for suppression alone and also data perturbation. Moreover, the algorithm is able to adjust by itself the main parameter z as data arrives to be more flexible if volume and type of data significantly change over time. The result are tested and evaluated on different use cases to provide a context as comprehensive as possible of the possible merits and limitations.