Differential Privacy Data Protection Method Based on Clustering
| Title | Differential Privacy Data Protection Method Based on Clustering |
| Publication Type | Conference Paper |
| Year of Publication | 2017 |
| Authors | Li-Xin, L., Yong-Shan, D., Jia-Yan, W. |
| Conference Name | 2017 International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery (CyberC) |
| Date Published | oct |
| ISBN Number | 978-1-5386-2209-4 |
| Keywords | categorical attribute, clustering, composability, data availability, data privacy, Differential privacy, differential privacy data protection method, distributed computing, Human Behavior, ICMD-DP method, information disclosure, information loss, insensitive clustering algorithm, insensitive clustering method, Knowledge discovery, mixed data, numerical attribute, pattern clustering, privacy protection, pubcrawl, Resiliency, risk reduction, Scalability |
| Abstract | To enhance privacy protection and improve data availability, a differential privacy data protection method ICMD-DP is proposed. Based on insensitive clustering algorithm, ICMD-DP performs differential privacy on the results of ICMD (insensitive clustering method for mixed data). The combination of clustering and differential privacy realizes the differentiation of query sensitivity from single record to group record. At the meanwhile, it reduces the risk of information loss and information disclosure. In addition, to satisfy the requirement of maintaining differential privacy for mixed data, ICMD-DP uses different methods to calculate the distance and centroid of categorical and numerical attributes. Finally, experiments are given to illustrate the availability of the method. |
| URL | https://ieeexplore.ieee.org/document/8250328 |
| DOI | 10.1109/CyberC.2017.15 |
| Citation Key | li-xin_differential_2017 |
- information loss
- Scalability
- risk reduction
- Resiliency
- pubcrawl
- privacy protection
- pattern clustering
- numerical attribute
- mixed data
- Knowledge Discovery
- insensitive clustering method
- insensitive clustering algorithm
- categorical attribute
- information disclosure
- ICMD-DP method
- Human behavior
- distributed computing
- differential privacy data protection method
- differential privacy
- data privacy
- data availability
- composability
- clustering