Data sanitisation techniques for transactional datasets using association rule hiding techniques
Nitin P. Jagtap, Krishankant P. Adhiya · 2023
The data mining and rule generation techniques are most popular due to the high utilisation of transactional data in e-commerce websites. Mining services require balanced data to generate accurate results, yet privacy concerns may lead consumers to supply false information. Several strategies based on random disturbance of database files have recently been presented to safeguard client privacy in data analysis. Every day, online businesses deal with thousands of transactions, leading to privacy concerns. Clustering algorithm concealing is a privacy approach that focuses on hiding confidential material produced by online department’s shops, Face book information, and other sources. These strategies are used to detect sensitive rules and offer privacy to sensitive rules, resulting in Lost and Ghost rules. Techniques created so far have failed to provide improved results. This paper proposed a rule hiding technique for hiding sensitive information from generated rules by hash base apriori. First, we implement hash-based apriori to generate the association rules and various feature selection techniques has been used for hiding the rules. The inexpensive experimental analysis evaluates both rules from DB’ rules and DB rules and evaluates the data quality and data loss. The proposed system can reduce 0.12% data loss on an adult dataset.