A Machine Learning Approach for Inference Detection in Databases

摘要：

A major threat attributed towards a database system is inference attack. There are well designed access control mechanisms that are used to control the access of sensitive information in the database based on the user identity. But, such methods does not protect against the inference attack, whereby an adversary poses a sequence of valid queries and uses the results of such queries to determine some sensitive information. In this paper, we propose a machine learning algorithm to control such inference attacks in databases by detecting the inference condition before the query is processed. The algorithm maintains the relationship among different attributes of data objects and incrementally improvises these relationships as more and more queries are posed. The algorithm uses users present query and a log of past queries to update the degree of maliciousness associated with a particular user. The decision of allowing the current query is based on the degree of maliciousness of the user and the relationship among the attributes, the user has queried for.

关键词： Machine learning database inference inference detection database security

作者: Saurav Tiwari D.V.L.N. Somayajulu Neeraj Shukla Niket Kumar Singh

作者单位: Department of Computer Science and Engineering National Institute of Technology Warangal, India

会议类型: 国际会议

会议名称: 2011 International Conference on Database and Data Mining(ICDDM 2011)(2011年数据库和数据挖掘国际会议)

会议地点: 三亚

会议语种:英文

页码: 255-259

在线出版日期: 2011-03-25（万方平台首次上网日期，不代表论文的发表时间）