An Evaluation Method for Nodes in Multiple Dynamic Networks

With the rapid development of internet and information technologies, ranking methods only based on simple citation (link) networks are already insufficient for the growing requirement of information extraction and evaluation. We propose a method for evaluating the influence and importance of nodes that are contained in multiple dynamic networks. This method combines a modified PageRank algorithm and modern data mining techniques such as a community detection algorithm and artificial neural networks (Self-Organizing Maps, SOMs). A citation network from the e-print Arxiv HEP-TH (high energy physics theory) is processed to serve as main dataset in the experiment, as well as data from social networks such as Facebook and Twitter. Some clustering results and visualizations, which indicate the influential trends (through time) of both individual nodes in a network and the network itself, are produced using different configurations.