Data Mining: A Threat to Privacy

Abstract

Data mining has emerged in the twenty first century as a crucial field that has a vast number of important real world applications. Use of data mining expands from business use cases to solving crimes such as credit card frauds. Various techniques in data mining is used to find hidden relationships and patterns from data sources that keep growing exponentially with time. With the advancement of technology, these data sources do not reside in a single central location. Rather the sources are dispersed all over the world. Data mining techniques that were used for mining single data repository is now modified to be used across a distributed environment. This enhances the power of data mining as well as the clarity of knowledge obtained because simultaneous mining of distributed data sources brings out a vast collection of hidden relationships that cannot be found in a single repository. Although distributed data mining applications brings in more results, it poses a great risk on the involved parties because of security concerns. Distributed mining is done using communication channels and the risk of security being compromised is increasing as the number of data sources increase. Since data mining applications are primarily used for finding hidden relationships among data, these findings itself can be threatening to the owners of the data sources since they might contain sensitive information that should not be exposed. Privacy preserving of data mining is crucial when using specially distributed methods because the privacy can be easily breached using very advanced techniques. This research finds a solution in protecting the privacy of the involved parties when mining association rules in a set of distributed databases. Association rule mining finds hidden relationships among data and the main focus of the research is to hide the rules of one party from the other parties as well as outside malicious intruders. By using the method introduced in this research, only the global result will be known to everyone and the privacy of each and every involved stakeholder is protected. This method not only preserves the privacy, but also protects the integrity of the final outcome of association rule mining.