Evaluating the impact of k-anonymization on the inference of interaction networks

We address the publication of a large academic information dataset while ensuring privacy. We evaluate anonymization techniques achieving the intended protection, while retaining the utility of the anonymized data. The published data can help to infer behaviors and study interaction patterns in an academic population. These could subsequently be used to improve the planning of campus life, such as defining cafeteria opening hours or assessing student performance. Moreover, the nature of academic data is such that many implicit social interaction networks can be derived from available datasets, either anonymized or not, raising the need for researching how anonymity can be assessed in this setting. Hence we quantify the impact of anonymization techniques over data utility and the impact of anonymization on behavioural patterns analysis.