Deep Cross-Network Alignment with Anchor Node Pair Diverse Local Structure

Network alignment (NA) offers a comprehensive way to build associations between different networks by identifying shared nodes. While the majority of current NA methods rely on topological consistency assumption, which posits that nodes across typically have similar local structures or neighbors, we argue anchor nodes, play pivotal role in NA, face more challenging scenario is often overlooked. In this paper, conduct extensive statistical analysis investigate connection status labeled node pairs and categorize them into four situations. Based our analysis, propose an end-to-end network framework uses representations as distribution rather than point vector better handle structural diversity networks. To mitigate influence specific introduce mask mechanism during representation learning process. addition, utilize meta-learning generalize learned information other pairs. Finally, perform experiments both real-world synthetic datasets confirm efficacy proposed method. The experimental results demonstrate model outperforms state-of-the-art significantly.