Identifying Influential Nodes in Complex Networks Based on Local Effective Distance

Abstract

With the rapid development of Internet technology, the social network has gradually become an indispensable platform for users to release information, obtain information, and share information. Users are not only receivers of information, but also publishers and disseminators of information. How to select a certain number of users to use their influence to achieve the maximum dissemination of information has become a hot topic at home and abroad. Rapid and accurate identification of influential nodes in the network is of great practical significance, such as the rapid dissemination, suppression of social network information, and the smooth operation of the network. Therefore, from the perspective of improving computational accuracy and efficiency, we propose an influential node identification method based on effective distance, named KDEC. By quantifying the effective distance between nodes and combining the position of the node in the network and its local structure, the influence of the node in the network is obtained, which is used as an indicator to evaluate the influence of the node. Through experimental analysis of a lot of real-world networks, the results show that the method can quickly and accurately identify the influential nodes in the network, and is better than some classical algorithms and some recently proposed algorithms.

Epidemic Model-based Network Influential Node Ranking Methods: A Ranking Rationality Perspective

“…Here, variable 𝑘 can be adjusted by the size of networks. Thus, the number of same nodes among the top-k nodes in new proposed methods and classic methods can show the similarity of information considered and illustrate the credibility of the new proposed methods [124,133]. For example, Zhang et al [133] applied Closeness Centrality [101], PageRank [93], Eigenvector centrality [27], H-index and LeaderRank [82] as ranking standards.…”

“…Thus, the number of same nodes among the top-k nodes in new proposed methods and classic methods can show the similarity of information considered and illustrate the credibility of the new proposed methods [124,133]. For example, Zhang et al [133] applied Closeness Centrality [101], PageRank [93], Eigenvector centrality [27], H-index and LeaderRank [82] as ranking standards. The more same nodes between the top-k node set or ranking list of the proposed INRM and the results obtained by these standards are, the better the ability of the INRM to assess ranking credibility is.…”

“…But in most cases, information will propagate randomly between multiple nodes. A classical example is in Yeast network, the transmission of neurotransmitters that randomly select their receptors [3,7,20,42,47,68,105,122,132,133,140,150,151]. Other scenarios like mutations in Protein-Protein Interactions or Human protein networks are random.…”

“…When they infect other proteins or cells, the nutrients or molecules would not be copied during the transmission process [105,133]. Therefore, similar to the case in technological networks, the information in biological networks is usually some biological substance, so its mode of spread is transfer, and there are usually many "receptors" in each information transmission round in the biological network.…”

“…In most cases, there are the same nodes in the top-k nodes. Some studies consider that it is not meaningful to compare the same nodes [107,133]. Therefore, the strategy is that the same nodes existing in the top-k nodes will not be considered, but the empty parts of the top-k nodes are filled by other nodes that follow sequentially in the ranking lists.…”