Open source software (OSS) has become a mainstream in software development, utilizing a global information infrastructure. OSS is a complicated social process to understand. OSS is a multi-faceted phenomenon including code, a licensing structure, a community, development best practices, a method of diffusion. However, the current OSS collaborative researches place too much emphasis on collaborative behaviors, but ignore the study on collaborative process. By using the social network theory to abstract collaborative network topology, this paper proposes a method for constructing social network model, which considers both the contact relationship and level of collaboration between collaborators. Based on the definition of three types of contact behavior, this paper presents an approach to measuring the contact relationship intensity. Based on introducing and improving TF-IDF (term frequency–inverse document frequency), this paper presents the methods for calculating tag weights and work similarity between collaborators. Finally, by evaluating the model using data from the OSS website www.Codeplex .com, we verify that our model outperforms conventional models in both describing and forecasting collaborative behavior.