Combat network link prediction based on embedding learning

Link prediction of combat networks is significant military value for precisely identifying the vital infrastructure enemy target and optimizing operational plan our side. Due to profound uncertainty in battleground circumstances, acquired topological information opponent network always presents sparse characteristics. To solve this problem, a novel approach named embedding based link (NECLP) put forward predict missing links networks. First, node techniques are presented preserve as much possible using low-dimensional space. Then, we solution algorithm between on similarity. Last, massive experiments carried out real-world case verify validity practicality proposed NECLP. This paper compares six baseline methods, experimental results show that NECLP has outstanding performance substantially outperforms methods.