Relation Extraction on the Wargame based on Pre-trained Models

It is of great significance to study the data of military chess deduction to improve the level of machines in military chess deduction software. Currently, the research on the data of wargame has not yet started. Data mining is carried out around the military chess deduction data. This paper mainly focuses on the relation extraction task. In this paper, relation extraction is regarded as a sequence tagging task, and a weighted fusion method based on pre-trained model is proposed. The experimental corpus is used to verify the feasibility of the proposed method. The experimental F1 value is 0.79, which verifies the feasibility of the method. The research work of this paper has accumulated experience for the follow-up research on data mining of the wargaming.