Abstract
One challenging issue in information science, biological systems, and many other fields is determining the most central or relevant networked systems agents. These networks usually describe scenarios using nodes (objects) and edges (the objects' relations). The so-called standard centrality measures aim to solve this kind of challenge, ranking the nodes by their supposed relevance and elect the most relevant nodes. This problem becomes more challenging when one single network is not enough to depict the whole scenario. In these cases, we can work with multiplex networks characterized by a set of network layers, each describing interrelationships that can change depending on external factors, e.g., time. This paper proposes a new centrality measure, the Group-based Centrality for Undirected Multiplex Networks, to find the most relevant nodes in an undirected multiplex network. As a case study, we use a Brazilian corruption investigation known as the Car Wash Operation. Our proposed centrality outperforms well-known centrality methods such as betweenness, eigenvector, weighted degree, Multiplex PageRank, closeness, and cross-layer degree centrality.
Highlights
Imagine a large number of individuals interacting in a network
Considering the example and supposing that each book is modeled as a separate complex network, obtaining the ranking of the whole book collection, it would be necessary to add the networks that model each book as a single network and only use a standard centrality metric to obtain the most relevant nodes
The analysis cannot consider these individuals; Denounced: the prosecutor lodges a formal complaint and the individuals are relevant for our analysis; Defendant: there was the acceptance of the complaint, and the individual will go to trial for some crime related to Car Wash Operation
Summary
Imagine a large number of individuals (nodes) interacting in a network. Somewhat ranking these individuals is a challenge, given that those interactions can vary in a non-predictable way. Considering the example and supposing that each book is modeled as a separate complex network, obtaining the ranking of the whole book collection, it would be necessary to add the networks that model each book as a single network and only use a standard centrality metric to obtain the most relevant nodes. This approach is flawed since various networks can have singular characteristics that change with time.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
