Abstract: This paper mainly studies the characteristics of researchers and subjects in the scientific project collaboration network and mines the research teams hidden in the network. Firstly, the cooperative project network is constructed by calculating the cosine similarity between personnel projects according on the research project data set. Secondly, social network analysis indicators are used to analyze this network. Finally, by combining the cohesive subgroup analysis algorithm with the merging strategy, a cohesive subgroups algorithm based on merge (MCSA) is proposed to mine scientific research teams hidden in the collaboration network. In the experiments, the relationship ketween the cooperation degree of the cooperation network of the city science and technology find the age structure of the core scientific research persomel and the development trend of each discipline were analyzed. Compared with FN algorithm and LPA algorithm, experimental results show that the scientific research teams mined by MCSA algorithm proposed in this paper are more suitable to the actual situation.

Key words: scientific project collaboration, social network analysis, cohesion subgroup, scientific research team