ID3 optimization algorithm based on interestingness gain

Aimed at the backwards of the information gain in ID3, through the improvement of information gain on interests of the users, and based on the calculating specialties of information gain in ID 3, the article reduces the backwards of decision tree´s attribute dependency towards more value by decision tree optimization through twice information gain and optimized calculation. The experiment proves that: compared with the traditional method, the optimized ID3 is provided with high accuracy and counting speed. In addition the structure decision tree possesses the advantage of lower average of leaf tree.