Privacy Preserving Association Rule Mining of Mixed Partitioned Model in Distributed Database Environment

Abstract:---- A big challenge to researchers is, how efficiently, one can extract knowledge in an application from a large database using appropriate data mining techniques. A well known research data mining technique is association rule mining whose outcomes are being successfully used in various real life applications to take strategic decisions. But, in recent years, people wish to share their knowledge which is predicted from association rule mining to their collaborators to get mutual benefits. But the problem is, they do not want to give their sensitive data to the collaborators because disclosure of sensitive data may cause harm to themselves. Here, the main issue is, how to achieve the two goals that is sharing accurate non sensitive information while protecting sensitive information. To address this issue, researchers have found many efficient methods for finding privacy preserving association rule mining. To enhance the availability and reliability of the customer services, distributed database applications are well being used which can also find association rules at global level, called global association rules. In this paper, privacy preserving association rule mining at global level is addressed. A global database usually partitioned in different ways but basically in three common ways such as horizontal, vertical and mixed model is a combination of both. In mixed partitioned model, a global database can be partitioned first in horizontal then each horizontally partitioned database is further partitioned into vertical or vice versa or in any order with different levels. This paper presented two efficient methods for determining privacy preserving association rule mining for two common mixed models by adopting cryptographic techniques, Sign based secure sum concept and scalar product technique