A GA-based approach to hide sensitive high utility itemsets.

Area of Science:

• Computer Science
• Data Mining
• Cybersecurity

Background:

• High-utility itemset mining is crucial for business intelligence.
• Existing methods often compromise data privacy when dealing with sensitive information.
• Protecting sensitive high-risk data while enabling utility mining presents a significant challenge.

Purpose of the Study:

• To propose a privacy-preserving utility mining method using a Genetic Algorithm (GA).
• To effectively hide sensitive high-utility itemsets through strategic transaction insertion.
• To balance information utility for data demanders with robust data protection.

Main Methods:

• A GA-based approach is employed for privacy-preserving utility mining.
• Transactions are strategically inserted into the database to obscure sensitive high-utility itemsets.
• A flexible evaluation function with three weighted factors assesses transaction suitability for insertion.
• Downward closure property and prelarge concept are utilized to optimize database rescanning.

Main Results:

• The proposed method successfully hides sensitive high-utility itemsets.
• Low information loss is maintained, ensuring data utility for legitimate users.
• High-risk information within the database is effectively protected.
• Reduced database rescanning costs and accelerated chromosome evaluation were observed.

Conclusions:

• The GA-based privacy-preserving utility mining method offers an effective solution for data protection.
• The approach successfully balances data utility and privacy preservation.
• Optimization techniques significantly improve the efficiency of the mining process.