Analysis Of Assessment Methods Based On Privacy Preserving For Data Mining

Data mining is to dig previously unknown mode out from the large amounts ofdata.There are many companies around the world often tackle with much datasystematically,and depend on digging out large data set to enhance their power. Whileone of the defects is these data sets include sensitive personal information,and suchprocess of mining will easily expose sensitive personal information to others. So oneof the aim is to ensure the safety of private data during data mining. Privacypreserving data mining is a special activity during data mining.Compare withextracted knowledge, the privacy of data is also important,due the its effects in theprotection of privacy. Data mining which is based on privacy protection is the way tofinding a balance between the privacy of original data and the accurate result ofdata mining.At this stage at home and abroad, many researchers have developed a variety ofprivacy-based data mining technology. Privacy preserving data mining algorithms aredivided into data encryption and data disturb technology, Mainly to protect theoriginal data conversion, and then data mining after joining the interference data sets,this method not only can excavate useful information but avoid the disclosure ofsensitive information.In privacy preserving data mining, there exist a mutual restraint betweenaccuracy and security, the improvement of one performance indicator will lead toanother’s decreasing, and these two indicators cannot reach the top at the same time.Thus,in order to obtain the best performance, we need adjust these two parameter.Privacy of data mining is under the condition that the user’s privacy will not bedamaged, and the useful moed can still be find. The better level of the protection, theworse the effect of data mining. Hence, it is essential for researchers to find a method to achieve the goal, which can both have a good accuracy and avoid privacy’sleakage. At present, the method of privacy assessment is to use convert data formto protect information, however, such method is usually measuring the privacy effectvia interval distance with reducing the number of distance to represent how muchprivacy is missing. But it will damage the efficiency of the data control and the resultof mining algorithms through this way. The majority of researchers are using thismethod to measure the relationship between the loss of information and privacy,which is a measure between the accuracy and security.The level of privacy protection is usually measured by the difference between theoriginal data and the data after interference. This difference is presented in theindicators through different pattern, such as distance, the greater difference withoriginal data, the better the protection of privacy. The accuracy of data is measuredby the similarity between the result of data mining after interference and original data.Thus, we can know that the more similar between the data after interference andbefore, the more accuracy the digging is.This paper is based on privacy preserving data mining, having a further study onprivacy assessment, and giving a analysis on the exited assessment methods in detail.In order to improve the performance of privacy, we obtained two groups of points withthe same distance from the interference with the original records point throughrotation after interference, and we have a data mining again after interference withthe two data sets. The experiments indicates that there exits some difference betweenthe two data sets after adding interference.The work includes three aspects as listed below:1.The introduction of privacy preserving methods.A study on the classical waysof data interference of privacy protection,and achieved a typical algorithm of additiveinterference.2.Study the typical performance indicators on privacy, and analyses somerelative theories. At the same time, analyses the theory of VD index and principle indetail.3.The accuracy of data mining will be different when discussed under the same level of privacy protection that based on the evaluation of the degree of privacy policy.In the case of the same degree of privacy protection VD indicators, this paper madethe following conculsions by analying the accuracy of data mining: we can guaranteethe same degree of privacy protection to improve the accuracy of data mining. thispaper processed the algorithm of interfered addition with the experimental data,and guarantee the same degree of privacy protection to improve the mining accuracy..