Research on key node identification method based on multi-attribute fusion in complex networks

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

Accurate assessment of node importance in complex networks is crucial for enhancing network robustness and security, for instance, by protecting critical data nodes or reinforcing core hubs. However, existing importance metrics often struggle to simultaneously accommodate local connectivity, neighborhood influence and global mediation effects. Moreover, they frequently yield similar importance scores for a large number of nodes, resulting in insufficient discriminability and failing to effectively identify key nodes. In this paper, we propose the CRITIC-based Multi-Attribute Fusion Indicator (CR-MAFI) within the CRITIC (Criteria Importance Through Intercriteria Correlation) framework. CR-MAFI integrates degree centrality, degree and neighborhood information centrality, and betweenness centrality, and employs Shannon entropy and inter-criterion conflict to derive objective weights, thereby enhancing the dominance of high-discriminability metrics in the fusion process. By integrating network structural information from multiple perspectives, the proposed method provides a more comprehensive assessment of node importance. Experiments are conducted on four types of real-world networks, including the aviation network USAir97, social network fb-pages-food, criminal collaboration network crime-moreno and power network 1138_bus. The results show that CR-MAFI achieves stable and relatively good performance in both simulated attack and spreading experiments, exhibits reliable damage effectiveness and spreading influence, and demonstrates reliable robustness and applicability across different types of networks and task scenarios. Analysis of the score distribution further reveals pronounced skewness and positive excess kurtosis across all tested networks, confirming that CR-MAFI consistently reproduces a score distribution that discriminates a small set of nodes.