Abstract: In order to better organize the evidence data of the South China Sea, improve the retrieval function of the South China Sea Think Tank, and facilitate researchers to conduct in-depth research on the South China Sea issue, this paper combines the ethnic retrieval of the Chinese Library Classification and the feature retrieval of the faceted classification, constructs the South China Sea document classification, applies the BERT pre-training model to realize the automatic classification of the fine-grained multi-label South China Sea evidence data, and uses the database technology to divide the evidence data of the South China Sea into 12 facets for grouping and retrieval. Accordingly, each piece of South China Sea evidence data is given a unique classification number and category name, which is formally expressed as “the data classification of South China Sea evidence data = evidence source classification + evidence data system classification + evidence data facet classification”. From a more fine-grained perspective, data classification is realized. Compared with the classification of broad-type document carriers, data classification is more conducive to data mining, to the identification of potential relationship between data, and to the discovery of new research perspectives, which are the foundation of scientific research and innovation.

Key words: