Research On Modeling Of Network Evolution And Knowledge Dissemination In A Scientific Collaboration Network

Recently the scientific collaboration networks have been extensively investigated in academia. In a scientific collaboration network, the nodes are the individual scientists; and the edges are the collaborations between the scientists, for example, the coauthorship relationships between them. In this thesis, the agent-based modeling approach is adopted to investigate such scientific collaboration networks by combining the aspects of the network evolution and the knowledge dissemination.Guimera et. al. provided a new research prospective for the scientific collaboration networks. In their model, scientists collaborate by organizing teams. Through agent-based computer simulation, they investigated the macro-level evolution of the entire scientific collaboration network by examining the local mechanism of team assembly. Furthermore, they compare the simulation result with the real-world data to test the effectiveness of the model. Besides the concrete results drawn from their work, Guimera and his colleagues have remarkable contribution to the study of scientific collaboration networks in that they proposed a promising way of bottom-top modeling to study the scientific collaboration networks. However, the proposed model itself is oversimplified to some extent and the mutual influences between knowledge dissemination and the network evolution is not taken into account in their model.In order to overcome the limitations of Guimera et al.'s work, in this thesis a more general model to study the evolution of team-assembly-based scientific collaboration networks is proposed. Consequently the model proposed by Guimera et al. can be regarded as a specific case of the proposed general model. Following the proposed model, this thesis suggests a research framework, which is based on the agent-based and bottom-up modeling approach, to investigate how the micro-or team-level models affect the macro-level properties of the scientific network that is made up of the teams. With this research framework, the dynamics of the scientific collaboration network and the dissemination of knowledge within the same network can be studied in a unified modeling framework.To facilitate the development of agent-based models for investigating the team-assembly-based scientific collaboration networks, an agent-based computer simulation platform is subsequently developed. The platform, which is based on a third-party Java-based class library to investigate networks and is able to make some detailed data analysis by integrating a few other third-party toolkits, is composed of three parts:the basic class library, the utility package and the user API.To test the usefulness of the proposed modeling framework and the corresponding computer simulation platform, this thesis uses the developed platform to implement and study the network evolution model proposed by Guimera et. al. as well as its extensions from three different perspectives. What's more, a comprehensive model that combines the scientific collaboration and knowledge dissemination is also conceptualized upon the proposed modeling framework and implemented by using the developed simulation platform.The key contributions of this thesis can be summarized as follows.Firstly, a modeling framework or "meta-model" that integrates the network evolution and knowledge transfer aspects is proposed. This framework is based on the bottom-top modeling approach of Agent-based Modeling. With the proposed modeling framework we can study the influences of the micro-level team formation mechanisms and knowledge exchange mechanisms on the evolution of and the knowledge dissemination within the whole network.Secondly, based on the suggested modeling framework, a computer simulation experimental platform is proposed for the agent-based modeling of scientific collaboration networks to facilitate the development of the simulation models and the analysis of simulation results.Finally, the usefulness of the proposed platform is validated by implementing a series of actual models by using the platform. This part of the work, on the other hand, contribute directly to propel the further investigations on the co-evolution of the scientific collaboration network and of the knowledge system embedded in the same network by using a unified modeling framework.