Abstract: Enterprises need a large number of new product development projects to gain competitive advantages, so knowledge diffusion in the process of new product development is particularly important. The paper focuses on the knowledge diffusion among multiple projects in the program. The program is a group of interrelated and coordinated among projects. Due to the dependency relationship between projects, the success of a project will enable the accumulated knowledge to be transferred to other projects in the program. The transfer of knowledge from one project to another is called knowledge diffusion. Specifically, knowledge diffusion means that the knowledge (including explicit and implicit knowledge) acquired by the research & development (R&D) team in one project can be transferred to other projects, thus driving the success of a series of other projects in the program. In all, the enterprise can continuously improve their R&D capabilities to better develop future products by transferring knowledge. Therefore, knowledge diffusion is conducive to improving success probability of the whole project program and promoting the sustainable development of enterprise. So, it is of great significance to analyze knowledge diffusion among projects.
To solve the problem of knowledge diffusion in the program, the paper proposes a quantitative analysis method based on the SIR (Susceptible Infected Recovered) model, so as to measure the efficiency of knowledge diffusion more accurately. Firstly, we build the communication network between teams, which is a directed weighted graph constructed with the project as the node and the communication frequency between teams as the edge. The direction of “edge” reflects the communication between teams, and the value (weight) of “edge” represents the communication frequency between teams. Then we use the Dependency Structure Matrix (DSM) and “tie strength” to measure the probability of knowledge diffusion among teams. The DSM is the form of matrix to represent the communication network. In the communication DSM, the “columns” in the matrix represent the project team's communication to other teams (i.e., output communication), the “rows” represent the project team's communication from other teams (i.e., input communication), and the non-diagonal numbers represent the communication frequency between project teams. The tie strength can be calculated by the proportion of the dependency between node i and j in all sending and receiving relations of the node i. Then, based on the characteristics of knowledge diffusion in the program, the paper defines the project teams as “knowledge diffuser”, “knowledge receiver”, “potential knowledge diffuser” and “knowledge immunity”. Knowledge diffuser refers to a project team that has mastered a certain type of knowledge and has the ability to transfer knowledge. Knowledge receiver refers to a project team that has the ability to accept knowledge. Potential knowledge diffuser refers to a project team that has acquired a certain kind of knowledge, but does not have the ability to transfer it. Knowledge immunity refers to a project team that temporarily withholds or does not diffuse knowledge from other project teams. We measure the probability of projects in different states and project knowledge level. Further, the SIR model is used to build the dynamic model of knowledge diffusion in the program network, and the dynamic propagation process of knowledge is simulated and analyzed. Finally, a certain program is taken as an example toverify the effectiveness of the model and method proposed in the paper. And, we establish the measurement indicators to analyze the efficiency of knowledge diffusion. The results show that in the process of multi-project research and development, the following measures can be taken to promote new knowledge or technology more efficiently and improve the performance of knowledge diffusion, such as improving the communication frequency between teams, strengthening the knowledge diffusion willingness of knowledge diffuser and the acceptance degree of knowledge receiver to new knowledge and improveing the knowledge diffusion ability of the project team and the ability to accept new knowledge again. In the future studies, we can further analyze the dynamic propagation mechanism of knowledge in the program and the influence of the variation of propagation probability on the performance of program. In addition, how to measure the effect and efficiency of knowledge diffusion in the program from other perspectives and what other factors affect the dynamic transmission of knowledge between projects are also worth further research.

Key words: project management, program network, knowledge diffusion, SIR model, dependency structure matrix

摘要： 针对知识在研发项目间的动态传播问题,本文基于传染病(SIR)模型构建了知识的动态传播模型。首先,在项目团队间沟通网络构建的基础上,采用依赖结构矩阵(DSM)方法利用节点间的关联强度测度项目间的知识传播概率;然后,依据对项目间知识传播过程的分析,将知识传播过程中的项目团队界定为四种状态,并分析了知识传播导致的项目在各状态间的转变;进一步,分别测度知识传播导致的项目处于不同状态的概率和项目的知识水平;最后,利用传染病模型构建项目集网络中知识传播的动力学模型,对知识的动态传播过程进行求解与仿真分析。

关键词: 项目管理, 项目集网络, 知识传播, 传染病模型, 依赖结构矩阵(DSM)