Abstract: The mining of successive examples is a central part in numerous information mining undertakings. A lot of research on this issue has prompted the excessive need of efficient and scalable algorithms for mining frequent patterns. Meanwhile, discharging these examples is posturing worries on the protection personal data of the clients participating. In this proposition, we examine the mining of successive examples in a protection saving setting. We propose an approach for differential private frequent item-set mining based on LCM algorithm; we refer it as P-LCM algorithm. P-LCM is extended version on PFP growth algorithm which basically works in two phases as pre-processing and mining phase. The first phase being the pre-processing phase it needs to be performed only once and smart transaction splitting method is used in this phase for improving utility as well as privacy trade off. Second phase limits the information loss caused by splitting as well as reduces the amount of noise added during mining process. LCM is an algorithm which finds all frequent item sets in polynomial time per item set. The closed item-sets obtained earlier are not stored in memory. The computational experiments on real world and synthetic databases exhibit the fact that in comparison to the performance of previous algorithms, our algorithms are faster and also maintain high degree of privacy, high utility and high time efficiency simultaneously.
Keywords: Differential Privacy, Frequent Item-set Mining, Transaction Splitting.