The Algorithm Of Decision Tree And Association Rules For K-Anonymity Data

The problem of data utility is a challenging problem for k-anonymity data. In 1998, L.Sweedy raised the k-anonymity privacy protection model, generalized quasi-identifier. Although it can protect the privacy data of users, greatly reducing the utility of data. In order to protect the privacy of users and to improve the utility of data in the same time. The researcher improved the k-anonymity privacy protection model. But it is constrained by the privacy protection requirements. Even if the optimal k-anonymity algorithm, it can not produce the accurate data. Therefore, improving the algorithm of k-anonymity is efficient in some extent. It can not solve the utility problem.Data mining is one of problems for the utility of anonymized data under the k-anonymity privacy protection model. Through analysis, finding that it is need to generalize both the quasi-identifier attribute values in the k- anonymity table and the node except leaf of the decision tree in the private table. According to this correspondence, we propose a decision tree algorithm based on k-anonymity. The algorithm accept the k-anonymity table as input, directly. To avoid the classical algorithm data preparation work before running.Experimental results show that there are significantly improved.Decision tree is an important data mining algorithm. The main purpose of the algorithm is used to classification and prediction. The article mentioned four circumstances about the classification of k-anonymity data. They are:(1) Classifying anonymized data using data mining models built on anonymized data. (2) Classifying original data using data mining models built on anonymized data. (3) Classifying anonymized data using data mining models built on original data.(4) Distributed data mining using anonymized data. Improved decision tree algorithm is the use of anonymous data as input, namely the use of anonymous data to establish the model. In the text, introducing the first two circumstance.Association rules is an important research branch in data mining, it used to find interesting correlation or related links between itemsets in amount data. According to the rule set can be covered by the rule abstraction layer into single association rules and multi-level association rules. The multi-level association rules is that the rules of the items is the accurate generalization in the table, and in many applications at the bottom or the original level of abstraction between data items is difficult to find a strong association rules. In general, it need to mine the multi-association rules. In this sense, they are common between the multi-level association rule mining process and k-anonymous generalization. Because the k-anonymity data is a special kind of uncertain data, in the article impoving the classic apriori algorithm, making it suitable for k-anonymity privacy protection model. Experimental results show that the algorithm proposed in this paper effective...