Anonymity and data mining

Atzori M., Bonchi F., Giannotti F., Pedreschi D.

Database Applications  Data mining  Frequent Patterns Mining  Data Privacy  Algorithms 

Improving trust in the knowledge society is a key requirement for its development. Privacy-awareness, if addressed at a technical level and acknowledged by regulations and social norms, may foster social acceptance and dissemination of new emerging knowledge-based applications. This is true of data mining, which is aimed at learning patterns, models and trends that hold across a collection of data. While the potential benefits of data mining are clear, it is also clear that the analysis of personal sensitive data arouses concerns about citizen's privacy, confidentiality and freedom. In this paper we focus on individual privacy, or anonymity, from a data mining perspective. It is generally believed that data mining results do not violate the anonymity of the individuals recorded in the source database. In fact, data mining models and patterns, in order to ensure a required statistical significance, represent a large number of individuals and thus conceal individual identities: this is the case of the minimum support threshold in association rule mining. In this paper we show that this belief is ill-founded. By shifting the concept of k-anonymity from the source data to the extracted patterns, we formally characterize the notion of a threat to anonymity in the context of pattern discovery, and provide a methodology to efficiently and effectively identify all possible such threats that might arise from the disclosure of a set of extracted patterns. On this basis, we obtain a formal and effective notion of privacy protection that allows the disclosure of the extracted knowledge together with the proof that it does not violate the anonymity of the individuals in the source database.

Source: Computer systems science and engineering 20 (2005): 369–376.

Publisher: Butterworth Scientific., Guildford, Regno Unito