Second-Order Neighbor Node-Driven Importance Ranking Algorithm

在复杂网络中，挖掘重要节点对精准推荐、交通管控、谣言控制和疾病遏制等应用至关重要。为此，本文提出一种局部信息驱动的节点重要性排序算法Leaky Noisy Integrate-and-Fire (LNIF)。该算法通过获取节点的二阶邻居信息计算节点重要性，首先通过信息量衡量节点重要性，然后叠加符合条件的邻居节点重要性，进一步区分节点重要程度，最终获得节点排序。为验证LNIF的有效性，在11个真实数据集和1个网络模型上进行了实验，采用平均度、网络效率、最大连通子图系数和SIR传播模型四个指标进行对比分析。实验结果表明，LNIF生成的节点序列在多个关键指标上表现显著优于现有基准方法。LNIF计算出的节点序列使网络平均度、网络效率和最大连通子图系数下降最快，表明其在优化网络结构和识别关键节点方面效率更高；在SIR传播模型中，LNIF的节点序列传播能力也优于其他五种算法，进一步证明了其在遏制和引导信息传播方面的潜力。In complex networks, identifying important nodes is crucial for applications like recommendation systems, traffic control, rumor containment, and disease transmission management. To address this, we propose the Leaky Noisy Integration-and-Fire (LNIF) algorithm, which ranks node importance using second-order neighbor information. LNIF first measures node importance by calculating information quantity, then refines the ranking by aggregating the importance of qualified neighbor nodes. Experiments on 11 real datasets and 1 network model, using metrics like average degree, network efficiency, maximum connectivity subgraph coefficient, and the SIR propagation model, demonstrate LNIF’s effectiveness. Results show that LNIF outperforms existing methods, with its node sequences causing the fastest declines in average degree, network efficiency, and maximum connected subgraph coefficient, highlighting its efficiency in optimizing network structure and identifying key nodes. Additionally, in the SIR model, LNIF’s node sequences exhibit superior propagation capabilities, further proving its potential in controlling and guiding information spread.