面向复杂场景战术知识的逻辑本体概率关系表征方法OA

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.