Abstract: 【Purpose/significance】Extracting and sorting out the elements of the South China Sea historical events is the basis for con?
structing the South China Sea historical events and telling the story of the South China Sea.【Method/process】First, summarizes the
particularity of the history of the South China Sea events, and then discusses the specific dimensions of the narrative of the South
China Sea, on this basis, the event element division criteria are defined to realize the standard modeling of the historical events in the
South China Sea, and then puts forward a combination of rules and deep learning event elements automatically extract method. Finally, the effectiveness and efficiency of the proposed method are verified by an empirical study on academic papers related to the South China Sea. 【Result/conclusion】 The results show that BERT+BiLSTM+CRF model performs better than other models, and the
Macro_F1 value reaches 87.73%. After optimizing the BERT+BiLSTM+CRF model with rule constraints, the Macro_F1 value reaches
88.76%, which achieves a good effect, and can quickly and effectively extract the event elements of historical events in the South
China Sea.【Innovation/limitation】Combined with the characteristics of historical events in the South China Sea, an automatic extrac?
tion method of event elements oriented to multi-dimensional narrative of the South China Sea is explored to realize the extraction of
various event elements in academic papers. Further generalization experiments need to be carried out on more literature data types.