Sub-Graph Degree-Based Bridge Centrality Algorithm

The design of centrality metric algorithms to distinguish influential nodes on networks has become a very attractive research interest lately. This recent development is due to the fact that detection of powerful nodes can be very helpful in various situations such as containment of epidemic spread and so on. Many useful centrality algorithms had been proposed in the past but some of them have inherent shortcomings. Some of these algorithms are designed to capture global or local network details but little efforts have been channeled to explore techniques based on sub-graph information. Therefore, a novel sub-graph degree-based bridge centrality (SDBC) algorithm to distinguish bridge nodes on networks is proposed in this work. A pivot node's sphere of influence is captured and this includes its 1-hop and 2-hop neighborhood with itself inclusive. A model is proposed to reduce links of all 2-hop neighbors to 1-hop links. The sub-graph degree distribution of all nodes under the pivot node's sphere of influence is computed and afterwards, the pivot node's bridging influence capacity is quantified using Dehmer's entropy model. The proposed algorithm is designed to capture the topological and strength of connections between nodes. In essence, a tuning parameter is used to automatically switch between unweighted and weighted networks, a feature that makes the algorithm unique. Additionally, SDBC does not use two factor computation unlike previous bridge node detection algorithms. Various experimental comparisons were carried out against some selected centrality metrics and SDBC performed well against the compared algorithms. The results prove that the proposed method is quite efficient in distinguishing bridge nodes on complex networks.