Journal of Science and Technology Metrics Jing Xu, Lina Wang 1 2 2020 10.6025/jstm/2020/1/2/71-76 https://www.dline.info/jstm/fulltext/v1n2/jstmv1n2_3.pdf Carrying out study on talent group composition can help us insight its present situation, and discover its potential problem, such as population aging, knowledge flow barrier, inbreeding, etc. An analysis model is proposed to characterize the talent group composition of research institute in three dimensions: age distribution, education experiences and collaboration pattern. Moreover, the analysis is based on objective data, and objective indicators. Age distribution may reflect the population aging problem whether or not; education experiences may reflect the in breeding problems; while collaboration pattern may reflect the knowledge flow problems. According to the indicator of each dimension, research institutes are divided into different types, the classification reveals different problems they may have. At last, an empirical study was conducted on two research institutes to prove the effectiveness of this model.