Analysis And Research Of Data Mining And Privacy Protection

With the rapid development of global Internet technology, network communic-ation technology and computer technology, global information network system has been the current sustainable infrastructure in all walks of life, network information system development have made tremendous contributions in the community. Since data mining technology mine from a mass of useful information out to people, so that data mining is an important step towards knowledge discovery. Data mining technology alone can easily lead to meaningless or misleading discovery mode, it is important to have a correct understanding of the application to use data mining. By association rule mining, we can get the valuable useful information hidden in huge amounts of data. With the development of network communication technology, data mining technology can not only provide people with knowledge and information but also expose the private information. Protection of private data or sensitive data in the data mining process can not be compromised while digging out the more accurate the results has become the emphasis and focus of the study of data mining technology.The main research work include the following:(1) A comprehensive description of the basic theory and application of the data mining technology.(2) Detail the work of association rule mining algorithm principle and method, and analyzes the typical Apriori algorithm, Because the candidate set which Apriori algorithm generate is too large, the algorithm must spend so much time dealing with candidate items, according to segmentation connectivity, the algorithm uses labels to record the number of matching item sets followed by this paragraph and has been optimized. According to the prior knowledge of k-item set is frequent item set, then all its (k-1)-items are a subset of frequent itemsets subtract the non-frequent itemsets in the k-items. reducing the number of 3-item sets items.Using space exchange time method, Boolean matrix is used to record the transaction records of database, only one scanning the database, so it greatly improves the executing efficiency of algorithm. (3) While mining division in the sequence, all records are owned by various participants, many participants do not want to disclose their private information at the same time as joint participants in the timing of each split sequences. We conver the timing rule to calculate the joint frequency of relatively problem into the comparison of the secret problem, and improve the existing algorithms and protocols by using the comparison of the secret agreement and homomorphic encryption agreement, then we put forward the new agreement.