Machine Learning Techniques as Mechanisms for Data Protection and Privacy

The exponential use of the Internet in the world has changed the way that data are collected, exchanged, and manipulated. Processing speeds, information volumes, and content relevance have changed rapidly to unimaginable levels. This is associated with the interest in creating personalized services based on the information available. Different organizations have data about people, representing a critical element, and it is estimated that up to 80% of all data stored in organizations can be classified as big data. The data stored by organizations usually comes from multiple sources (devices, people, organizations, among others), and these sources become the producers of the data. These data can be published or shared with other organizations or individuals, so the recipients of these publications become the consumers of the information. This research exposes problems in measuring the usefulness of data and relates them to the preservation of privacy in publishing. Machine learning models are developed to determine the risk of predicting sensitive attributes and as a means of verifying the utility of the data.