Distributed knowledge resources for the pig diseases diagnosis have the characteristics of heterogeneous, autonomy and difficult to be shared among different systems. Based on the web-based ontology modeling language, a descriptive model to describe the heterogeneous knowledge resources of the pig diseases diagnosis and a formal model to express the decision services of pig diseases are proposed in this work. Based on the two models, a complex decision task can be automatically divided into many dependent simple decision fragments. The relationships between different fragments, the mapping relationships between each fragment which can be dynamically adjusted according the divided fragments and its distributed knowledge resources can be formally described. The models form a mechanism for cooperative decision for pig diseases diagnosis. The method is proposed by achieving the balanced decomposition of decision task and intelligent schedule of knowledge resources. The method is verified in a pig diseases collaborative diagnosis system. The result shows that the method is superior to the traditional intelligent decision-making method.