Enhancing the Sharing and the Management of Personal Data in the Big Data Era

Nowadays, thanks to the digitalization of business processes and public administrations, many significant Big data collections are available. Users are direct suppliers of data when publishing contents on social networks. However, when using a service on the web, users must often provide their data, which will become property of the company running the service. To this end, users need to be aware of the privacy issues related to the management of their data, whereas companies need to ensure the protection of users' personal data, also according to new laws and regulations issued by governments. On the other hand, there exists the necessity not to limit the processing of data by companies and other public institutions. Thus, it is necessary to devise methods devoted to the identification of possible privacy threats during users' online activities, and to develop privatization strategies that possibly do not downgrade the significance of data. This dissertation provides experimental evidence of several threats for users when providing their personal data for accessing online services, aiming to increase their awareness, and it describes new methodologies and tools to support companies when processing personal data of their users. In particular, the proposed methodologies exploit data correlations expressed in terms of relaxed functional dependencies (RFDs) to define privatization strategies, aiming to safeguard user's privacy, and to detect malicious accounts in social networks. Finally, two automatic tools have been designed and implemented to help users better understand privacy threats during their online activities. [edited by Author]