This paper discusses the development of Chinese public libraries from both theoretical and practical aspects. This project proposes to study the problem of identifying and connecting Chinese entities based on the BERT-CRF (Bidirectional Encoder Representations from Transformers-Conditional Random Fields) model of translation-consistent random fields. Traditional methods of entity recognition and connection have many problems, such as insufficient feature expression, insufficient context information and limited association. In this project, BERT is combined with conditional random field to learn the text semantics, and the conditional random field model is used to extract the whole text to improve the accuracy of entity recognition and entity connection. Experiments show that the Chinese entity recognition and association algorithm based on BERT-CRF effectively improves the retrieval efficiency of database. The accuracy of entity association increases from 93.5 to 95.9%, and floats from 93.9 to 96%. The findings of this project will provide new avenues for entity identification and association studies.