Cleaning indirect identifiers is far more complex than direct identifiers; techniques such as differential privacy and k-anonymization can partially mitigate this but often at the cost of losing data utility.