An Efficient Privacy Preserving Association Rule Mining from Highly Secured Outsourced Databases

Abstract

Database outsourcing is becoming more commercial in the recent distributed and parallel
systems. This paper considers Association Rule Mining, Frequent Itemset Mining and Privacy
Preserving Mining. There is always a controversy between security and the flexibility. For
better mining approaches flexibility is more needed from the Database servers but it increases
its security risks on distributed network. In current trend of system setup on distributed clouds
the Database servers are separated from the service providing web servers. It also extends the
resource of the web servers where they can access more than one Database to analyze and
retrieve results. At this scenario the web server acts as the intermediate between the Database
servers and the client applications. It is responsibility of the web server to preserve privacy of
both client and the Database server. This paper concentrates on both client side and Database
server side privacy by introducing the algorithms No-Cache Rules Mining on client and
Encrypted Database Access on Database server to preserve both client and server privacy. By
internally it follows the traditional Association Rule Mining, Frequent Itemset Mining but in
different manner.