Datamining

Distributed Data Mining Protocols for Privacy: A Review of Some Recent Results⋆
Rebecca N. Wright1 and Zhiqiang Yang1 and Sheng Zhong2⋆⋆
1

2

Department of Computer Science, Stevens Institute of Technology, Hoboken, NJ 07030 USA Department of Computer Science & Engineering, State University of New York at Buffalo, Buffalo, NY 14260 USA

Abstract. With the rapid advance of the Internet, a large amount of sensitive data is collected, stored, and processed by different parties. Data mining is a powerful tool that can extract knowledge from large amounts of data. Generally, data mining requires that data be collected into a central site. However, privacy concerns may prevent different parties from sharing their data with others. Cryptography provides extremely powerful tools which enable data sharing while protecting data privacy. In this paper, we briefly survey four recently proposed cryptographic techniques for protecting data privacy in distributed settings. First, we describe a privacy-preserving technique for learning Bayesian networks from a dataset vertically partitioned between two parties. Then, we describe three privacy-preserving data mining techniques in a fully distributed setting where each customer holds a single data record of the database.

1

Introduction

The advances in networking, data storage, and data processing make it easy to collect data on a large scale. Data, including sensitive data, is generally stored by a number of entities, ranging from individuals and small businesses to national governments. By sensitive data, we mean the data that, if used improperly, can harm data subjects, data owners, data users, or other relevant parties. Data mining provides the power to extract useful knowledge from large amounts of data. However, most data mining techniques need to collect data from different parties; in many situations, privacy concerns may prevent different parties from sharing their data with others. An important technical... [continues]

Read full essay