Personalized privacy-preserving frequent itemset mining using randomized response.

Area of Science:

• Data Mining
• Privacy-Preserving Techniques
• Machine Learning

Background:

• Frequent itemset mining is crucial for association rule mining and pattern discovery in large datasets.
• Existing privacy-preserving methods for frequent itemset mining often apply uniform privacy levels, potentially reducing data utility.
• There is a growing need for methods that can handle varying privacy requirements across different data attributes.

Purpose of the Study:

• To introduce and address the problem of personalized privacy in frequent itemset mining.
• To propose a novel method for personalized privacy preservation in frequent itemset mining.
• To evaluate the effectiveness of the proposed method in balancing privacy and accuracy.

Main Methods:

• The study utilizes the randomized response technique to implement personalized privacy.
• A method is developed to assign different privacy levels to distinct attributes within the dataset.
• The proposed approach is applied to frequent itemset mining algorithms.

Main Results:

• The personalized privacy-preserving method demonstrates higher accuracy in frequent itemset mining compared to traditional methods with uniform privacy levels.
• Experimental results confirm the method's effectiveness in preserving personalized privacy while maintaining data mining performance.
• The approach successfully balances the need for data utility with granular privacy protection.

Conclusions:

• Personalized privacy levels in frequent itemset mining can lead to improved accuracy and data utility.
• The randomized response technique offers a viable solution for implementing personalized privacy in data mining.
• This work advances the field of privacy-preserving data mining by offering a more flexible and effective approach.