Efeitos da anonimização nos processos de mineração de dados

Abstract

This work proposes to apply anonymization and data mining techniques to a set of credit data, to verify data privacy, information security, and compliance with the General Data Protection Law (GDPL), in addition to the feasibility of continuing to apply decision-making processes to anonymized data. Data mining presents itself as a strategy to assist in decision-making on anonymized datasets. Bibliographical research on the GDPL, anonymization, and data mining was carried out to understand and conceptualize these themes. With the use of the amnesia software, anonymization techniques were applied to the credit dataset (original set), as data anonymization can be used to obtain privacy and compliance with protection laws and regulations of data. This strategy can be applied in a way that does not allow the re-identification of data and still contributes to information security and risk management. For each of the applied anonymization techniques, data mining was used with the help of the Waikato Environment for Knowledge Analysis (WEKA) tool and the accuracy values obtained. As a result, it appears that anonymization based on suppression and generalization ensured the necessary privacy, as well as did not prevent classification techniques based on decision trees from being used for adequate decision making.