A framework for evaluating privacy protection within data mining processes is presented in this paper. Experimental evaluations based on algorithms and a related testing portfolio are employed. The evaluation shows the findings on efficiency, scalability, data quality and selectability, and the privacy level reached for a relevant algorithm. The main privacy protection process in data mining involves hiding sensitive data. The authors search for new algorithms in this process, and, through the rule-hiding algorithms based on data fuzzification, they introduce the novelty with the insertion of rule confidence levels.
At the same time, the authors describe the framework that allows the use of different features of a privacy-preserving algorithm depending on evaluation criteria. At the end of the main text, an appendix follows presenting proofs of theorems on efficiency evaluation stated in the text. In this way, the authors help readers better understand the idea and evaluation process that they present. This paper is valuable reading for everyone coping with data mining technology when data sets are related to privacy protection terms.