Abstract

Text topic mining and visualization are the basis for clustering the topics， distinguishing front topics and hot topics. This paper constructs the LDA topic model based on Python language and researches topic mining， clustering and dynamic visualization，taking the metrology of Library and information science in 2017 as an example. In this model，parameter and parameter are estimated by Gibbs sampling，and the best topic number was determined by coherence scores. The topic mining based on the LDA model can well simulate the semantic information of the large corpus，and make the corpus not limited to the key words. The bubble bar graph of the topic-words can present the many-to-many dynamic relationships between the topic and words.

This work is licensed under a Creative Commons Attribution 4.0 License.