Research Field Discovery Based on Text Clustering

For the need of revealing the structure and dynamics of science, text mining technologies have been widely used in extracting technical intelligence from research literature. To analyze the status of research fields, this paper applies text clustering to Chinese national science research proposals. By using vector-space model to represent the proposals, an improved Newman fast clustering algorithm has been carried out to explore the clusters of each year, which indicate the yearly research fields. After that, the paper gives an insightful representation of the whole status of the research fields. Then the similarities of clusters of different years are calculated to discover the groups, which identify the major research fields and illustrate their stability and changes over the time. These analyses will be helpful to discover the development trends of basic research fields.