火力与指挥控制2026，Vol.51Issue(7)：41-48,8.DOI:10.3969/j.issn.1002-0640.2026.07.007
面向复杂场景战术知识的逻辑本体概率关系表征方法
A Probabilistic Relational Representation Method of Logical Ontology for Tactical Knowledge in Complex Adversarial Scenarios
摘要
Abstract
To address the mismatch between the probabilistic nature of temporal evolution and the determinism of logical ontologies,as well as the topological collapse problem in adversarial decision-making,a probabilistic relation geometric representation framework,probabilistic relation box(PR-Box),is proposed.This method maintains the rigid hyperrectangle boundaries of concepts to preserve the exclusivity of objective facts,while reconstructing tactical transition relations into continuous probability distributions.By computing the probability mass of the distribution falling into the target box,it achieves a mathematical alignment between Markov transitions and geometric ontologies.Experiments on large-scale adversarial datasets demonstrate that this method effectively preserves the underlying physical rule logic and significantly enhances the diversity and generalization capabilities of decision priors,validating its practical application value in complex adversarial decision-making.关键词
神经符号系统/εL++描述逻辑/概率几何嵌入/知识表示/时序演进Key words
neuro-symbolic systems/εL++description logic/probabilistic geometric embedding/knowledge representation/temporal evolution分类
信息技术与安全科学引用本文复制引用
王健,朱胤,王潆升,赵宇佳,申昊锴,吴心雨..面向复杂场景战术知识的逻辑本体概率关系表征方法[J].火力与指挥控制,2026,51(7):41-48,8.基金项目
国家级预先研究基金资助项目(315057210) （315057210）
