Privacy preserving classification algorithm based on Shamir secret sharing
Zou Han-bin · Jisuanji gongcheng yu sheji · 2010
According to Shamir secret sharing theory, a privacy preserving decision tree classification algorithm based on distributed environment is presented, to Classifying mine on the distributed environment and protect every part privacy.Firstly, the maximal information gain formula of the classification attribute is analyzed on the concentrating database decision tree, and the maximal information gain formula is deduced for the same classification attribute on the distributed data.Then, the encrypt principle of the Shamir shared is analyzed, and apply it to the maximal information gain formula of the decision tree classification attribute on the distributed database.A case planning is presented for the process of the privacy value summation.Finally, the experimental results show that this algorithm can effectively mine decision tree for isomorphism of distributed sample data sets to protect privacy.