The Method Of Data Discretization Based On Factor Space Theory And It Combine With The Set Subtraction And Rotation Calculation

In order to solve the problem of poor knowledge reasoning effectiveness when the set subtraction and rotation calculation is used to deal with continuous-factors.The data discretization algorithm based on factor space theory and it fusion with the set subtraction and rotation calculation were studied.The following work are mainly done.First,the equivalence between factor's conjunction,disjunction operation and Quotient set's sum,product operation was proved.Second,On the basis four artificial cognition principles of factor space theory,the factors' information conversion's ideas and methods were proposed.A new data discretization method suitable for the set subtraction and rotation calculation was constructed.Lastly,the decision-making process of the set subtraction and rotation calculation was unified expression by discrete-factors based on this.The effectiveness of the new fusion algorithm is verified on four different UCI datasets.The results showed that: The fusion algorithm significantly improves the generalization effect of continuous-factor experience knowledge.Compared with the decision tree of the same production reasoning algorithm,the empirical knowledge's generalization accuracy and the comprehensive model performance of the two algorithms are equivalent.But the empirical knowledge's traceability and interpretability of the set subtraction and rotation calculation are better than the decision tree.The empirical knowledge of the set subtraction and rotation calculation does not need Pruning treatment,and the empirical knowledge's structure of the set subtraction and rotation calculation is the fishbone diagram structure.So knowledge expression of the set subtraction and rotation calculation is more concise,and the application of the empirical knowledge's structure is simpler.