The key to obtain an effective knowledge base to improve decision quality and enhance organizational core competency is integrating the knowledge from different subjects and sources. We put forward a knowledge integration strategy based on culture algorithm framework. It encodes the knowledge uniformly and evolutes among the two phases of population space and belief space. Through the communication protocol established among the two spaces, an effective and concise knowledge base is obtained. The experiment shows that comparing with traditional genetic algorithm the model can classify the knowledge more precisely, reduce redundant knowledge and remove contradictory knowledge.