Abstract:

There is an increasingly urgent demand for intelligent command decision-making （CDM）， while recent achievements in China and abroad in military intelligence rarely touch this field. Lack of data， knowledge， and models has hindered the development of intelligence in this field. Based on the understanding of domain problems， the mentor-led idea of intelligent development in the field of command decision-making is innovatively put forward. According to the development of cutting-edge technologies such as knowledge graphs， reinforcement learning， and large-scale language models in recent years， two sets of mentor-led command decision-making application modes are constructed， namely “apprentice-support knowledge extraction” and “mentor-led wargaming”. In addition， the key technologies that need to be broken through are pointed out， which has certain guiding significance for breakthroughs in intelligent command decision-making.

Key words: command decision-making, human-machine integration, hybrid intelligence, knowledge extraction, wargaming, chatGPT

摘要：

指挥决策智能化需求日益迫切，而近年来国内外军事智能方面的成就较少涉及这一领域。缺数据、缺知识、缺模型的现状，导致该领域智能化发展相对滞后。基于对领域问题的理解，创新提出导师带徒的指挥决策领域智能化发展思路，并基于近年来知识图谱、强化学习、大型语言模型等前沿技术的发展，构想了“学徒式知识提炼”“授徒式博弈推演”2套导师带徒式指挥决策应用模式，并指出了其中需要突破的关键技术，对指挥决策智能化攻坚具有一定指导意义。

关键词: 指挥决策, 人机融合, 混合智能, 知识提炼, 博弈推演, 聊天机器人