In this paper, we mitigate a privacy threat to a large survey rating data set using an anonymity principle. We apply the flag matrix to formulate the problem, through which we provide
a greedy approach to anonymize rating data. Extensive experiments confirm that our technique
produces anonymized data sets that are highly useful and preserve the key statistical properties.