In this paper, a fuzzy model algorithm for a rule-base self-extraction and simplification is introduced. The method consistes of three steps: The first step is to classify the out-in space by constructing a fuzzy partition validity index, then the optimal number of dusters and, hence, the optimal number of rules are obtained; The second step is to construct the initial fuzzy system based on the optimal number of rules and neural networks; The third step is to get the function of computing similarity of fuzzy sets by fuzzy similarity analysis method. The similar fuzzy sets are merged to create a common fuzzy set in rule base based on threshold value. Thus a fuzzy system with interpretability and simplicity is obtained. At last, the fuzzy rules of productivity factor of China is extracted by the fuzzy system.