Ant Colony System Sanitization Approach to Hiding Sensitive Itemsets
Open Access
- 9 May 2017
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 5, 10024-10039
Abstract
In recent years, privacy-preserving data mining (PPDM) has received a lot of attention in the field of data mining research. While some sensitive information in databases cannot be revealed, PPDM can discover additional important knowledge and still hide critical information. There are different ways to approach this exhibited in previous research, which applied addition and deletion operations to adjust an original database in order to hide sensitive information. However, it is an NP-hard problem to find an appropriate set of transactions/itemsets for hiding sensitive information. In the past, evolutionary algorithms were developed to hide sensitive itemsets by building an appropriate database. Genetic-based algorithms and a particle swarm optimization-based algorithm, proposed in previous works, not only hide sensitive itemsets but also minimize the side effects of sanitization processes. In this paper, an ant colony system (ACS)-based algorithm called ACS2DT is proposed to decrease side effects and enhance the performance of the sanitization process. Each ant in the population will build a tour for each iteration and each tour indicates the deleted transactions in the original database. The proposed algorithm introduces a useful heuristic function to conduct each ant to select a suitable edge (transaction) for the current situation and also designs several termination conditions to stop the sanitization processes. The proposed heuristic function applies the pre-large concept to monitor side effects and calculates the degree of hiding information to adjust the selecting policy for deleted transactions. The experimental results show that the proposed ACS2DT algorithm performs better than the Greedy algorithm and other two evolutionary algorithms in terms of runtime, fail to be hidden, not to be hidden, not to be generated and database similarity on both real-world and synthetic data sets.Keywords
Funding Information
- U.S. Department of War (W911NF-17-1-0088, W911NF-16-1-0416)
- National Science Foundation (1710716, 1625677)
- Oak Ridge National Laboratory (4000144962)
- United Hall Foundation (1592)
This publication has 31 references indexed in Scilit:
- Sequential pattern mining -- approaches and algorithmsACM Computing Surveys, 2013
- Privacy preserving data mining: A noise addition framework using a novel clustering techniqueKnowledge-Based Systems, 2011
- Privacy preserving mining of association rulesInformation Systems, 2004
- State-of-the-art in privacy preserving data miningACM SIGMOD Record, 2004
- Tools for privacy preserving distributed data miningACM SIGKDD Explorations Newsletter, 2002
- The compact genetic algorithmIEEE Transactions on Evolutionary Computation, 1999
- Data clusteringACM Computing Surveys, 1999
- Ant colony system: a cooperative learning approach to the traveling salesman problemIEEE Transactions on Evolutionary Computation, 1997
- Data mining: an overview from a database perspectiveIEEE Transactions on Knowledge and Data Engineering, 1996
- Ant system: optimization by a colony of cooperating agentsIEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), 1996