Association rule hiding using cuckoo optimization algorithm. (1st December 2016)

• Record Type: Journal Article
• Authors: Afshari, Mahtab Hossein; Dehkordi, Mohammad Naderi; Akbari, Mehdi
• Is Part Of: Expert systems with applications. Volume 64(2016)
• Page Start: 340
• Page End: 351
• Publication Date: 2016-12-01
• Subjects: Data mining -- Privacy preserving data mining -- Association rule hiding -- Cuckoo optimization algorithm; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2016.08.005
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: We use Cuckoo Optimization Algorithm for hiding sensitive association rules. A preprocess is defined that causes speedy access to the optimal solution. Introducing three fitness functions with minimum side effects. An immigration algorithm is defined to escape from local optimums. For efficiency assessment, the algorithm is examined on real and synthetic data. Abstract: Privacy preserving data mining is a new research field that aims to protect the private information and avoid the leakage of this information during the data mining process. One of the techniques in this field is the Privacy Preserving Association Rule Mining which aims to hide sensitive association rules. Many different algorithms with particular approaches have so far been developed to reach this purpose. In this paper, a new and efficient approach has been introduced which benefits from the cuckoo optimization algorithm for the sensitive association rules hiding (COA4ARH). In this method the act of hiding is performed using the distortion technique. Further in this study three fitness functions are defined which makes it possible to achieve a solution with the fewest side effects. Introducing an efficient immigration function in this approach has improved its ability to escape from any local optimum. The efficiency of proposed approach was evaluated by conducting some experiments on different databases. The results of the execution of the proposed algorithm and three of the previous algorithms on different databases indicate that this algorithm has superior performance compared to other algorithms.