In order to excavate the core content and its problems of science popularization policy, ROSTCM, a kind of word frequency analysis software, is used to extract keywords, and co-word matrix of keywords and correlation and dissimilarity matrix are constructed; then Ucinet and SPSS are used to make cluster analysis, multi-dimensional scaling analysis and social network analysis of these keywords. In cluster analysis, the keywords of the popular science policy content are classified into 8 classes, which is respectively conducting training for youngsters and peasants, popular science resource sharing, infrastructure and fund input, etc. It shows that popular science policy covers a wide range of topics. In multidimensional scaling, popular science policy content is classified into 4 classes words, content of science popularization policy possess distinct focus points. By analyzing the keywords of the popular science policy content in network density, the study shows that there exist flaws in the content design of those policies. In a nutshell, the paper puts forward some suggestions which strengthen the hierarchy of policy content, and make policy content be more systematically conducted, etc.

visualization  /  popular science policy  /  content mining  /  co-word analysis