A Temporal Knowledge Graph Based Approach to Task Intelligent Inference
Inferring adversarial tasks is essential in military decision-making.However,some adversarial forces and weapons cannot be detected in real war,making the inference of their tasks challenging.A novel framework for task inference based on temporal knowledge graph reasoning is proposed.Firstly,a temporal task knowledge graph named TASK is constructed from simulated dataset.Information from both historical and concurrent task patterns are considered in the graph embedding process.The proposed model and baselines are tested on TASK and the public dataset YAGO.Experimental results show that the proposed model can yield better performance compared with baselines in task inference and all dynamic models perform better on TASK than static ones,which is likely due to the crucial role of temporal information in military task inference.
NETL
NSTL
万方数据
