Protecting Privacy in Large Datasets—First We Assess the Risk; Then We Fuzzy the Data