Abstract:
Today we are producing abundant data every day and need to extract useful information by applying data mining techniques. This is also known as Knowledge Discovery from Data (KDD). KDD cycle involves four key users including Data Provider, Data Collector, Data Miner and Decision maker. All these users have their own concerns, data privacy is the main concern of data provider, correct data collection is the concern of data collector, accurate results are the concern of data miner and decision maker.
Wide use of data mining techniques leads to data privacy issues and therefore need to address privacy preservation in KDD. Many techniques exist to cater this issue like Perturbation, Condensation, data hiding etc. Each tries to preserve privacy at maximum level but this leads to data loss resulting compromised data mining results. It has been observed that if privacy not compromised then accuracy of results is compromised hence data mining results get affected.
Most critical Quasi-identifier attributes need to identify in the data set as these can help attacker to identify individual and hence lead to privacy leak. Once identified such attributes, need to take appropriate measure and preserve privacy.
We are proposing a solution not only to preserve privacy with minimum accuracy loss which is concern of data provider but also taking care of concerns of other users as well by introducing trusted third party. Trusted third party will take care of the interests of users involved in KDD cycle from data provision to decision making.
