Abstract: The forecast of the technology progress probability through BNM, its implication of the technical and economic indicates the character of the technology progress, corresponds to the process of the technology innovation for the firms. At first, the thesis analyses the sets of the technology progress in each stage and their relations, inferences and judges the relevance of the pre-stage of the technology progress and the current stage, the current stage of the technology progress and next stage, the condition of the rational forecast of the technology progress is more enough. According the research, the firms select the technical diffusion, rational forecast of technology progress is feasible, is a proceed in an orderly way and step by step, from un-rational forecast to bounded rational, then to whole rational forecast. The sets of the probability of the technology progress have the character of RIP, the probability of each stage of the technology progress has relation only with the pre-stage, with no relation of the next stage. The intersection of the probability sets among two stages of the technology progress is larger, their relevance is more closer each other, the probability to advance higher stage is higher. The difference of the rational forecast of the technology progress is the necessary condition for optimization selection of the technology progress path, the result of rational forecast is more correct with shorter path of the technology progress. Empirical analysis confirms the nodes of the technology progress stage, draws DAG with patent citation, technical intense, general index, relation of the science and technical, then carries the analysis of the cause and diagnostic reasoning, and verifying the conclusion.

摘要： 运用贝叶斯网络模型（BNM）理性预测技术进步概率，既与技术创新流程相契合，又反映技术进步内在变动特征。文章首先分析各阶段技术进步概率集合以及集合之间相互关系，分析推理上阶段技术进步与当前阶段、当前阶段技术进步与下阶段之间的关联程度，依此理性预测技术进步概率。研究认为，厂商有意识地选择技术进步方向，理性预测其变动结果是可行的，是一个由不完全理性预测到比较理性预测、最后实现完全理性预测的循序渐进过程。技术进步概率集合符合集合动态分配律（RIP），两阶段技术进步之间概率集合交集越大，两者关联程度越高，技术向着更高阶段升级的概率就越大，技术进步预测更加准确。差异化的预测结果是厂商实现技术进步路径优化选择的必要条件，较短距离的技术进步预测结果更加准确。实证分析运用国内专利被引用次数、技术强度、科学与技术关联性等数据，绘制有向无环图（DAG），运用BNM进行因果分析和诊断推理，验证相关结论。