Abstract: 【Purpose/significance】Topic development hierarchy classification is the basic problem of information organization research,and it is also an important work for researchers and scientific research management departments to carry out research topic selection and subject service. Effective development hierarchy analysis and trend prediction of subject research topics can help relevant re⁃searchers and organizations to grasp the research situation of subject areas and make accurate scientific research decisions【. Method/process】In this paper,a topic development hierarchy analysis and trend prediction method of subject research was proposed by com⁃bining topic model,Sen's slope estimation method,Mann-Kendall method and exponential smoothing method.Firstly,on the basis of topic identification,two indexes of topic publication degree and topic citation degree were formed,and the topic development hierarchy of subject research was divided according to Boston Matrix.Then,the topic thermal intensity index was formed by integrating the num⁃ber of research topics published,cited and downloaded,and the exponential smoothing method was used to predict the future develop⁃ment trend of the research topics.【Result/conclusion】Experiments in the field of "smart library" show that the method proposed in this paper can classify and predict the trend of the research topics in an all-round and fine-grained way,and is applicable to the analy⁃sis of the research topics in other disciplines【Innovation/limitation】The method proposed in this paper is universal to other research
topics and provides a new perspective for dynamic intelligence analysis.However,the limitations of this paper lie in the need to improve the readability of topic modeling and further optimize the trend prediction method.