Identifying and Ranking Influential Spreaders In Complex Networks

Abstract

In this digital age, identifying Influential Users (IU) in Complex Networks (CN) has newlinebecome an urgent necessity, and it has evolved as a significant and fascinating research newlinesubject. For instance, influential users help endorse a product campaign or immunize newlinerumors in their network community. Only some users in any given real-world network newlinepossess the ability to accelerate or decelerate the information outbreak. Hence, identifying newlineand ranking these sets of users is an important factor for the same purpose. To find newlineout these influential spreaders, there are several classical and advanced techniques such newlineas Centrality Measure (CM), K-Shell (KS), etc. however, each of these methods has its newlinedrawbacks. Numerous researches on this topic depict the same and propose improved newlinemethodologies. However, most of the proposed techniques compute rankings based on newlineraw network structures alone. In a real-world scenario, numerous metrics lie in the network, newlinewhich act as the crucial factor for information cascade. So, to address this, in this newlinethesis, we propose improved ranking schemes that effectively utilize various underlying newlinekey network metrics and identify influential users who can make information cascade newlinemuch faster. newlineInitially, utilizing a Sentiment Weighted Page Ranking Algorithm (SWPR), the approach newlineoffers a methodology in which a user s influential rank is computed. Here we newlineutilize the well-renowned page rank algorithm score concept of calculating ranks based newlineupon its in-links and out-links; along with it, we extract and infuse nodes underlying newlinesentiment, which helps to narrow down nodes with the right spreading capability. The newlineranking is based on sentiment weighted page ranks value where the number of node newlineconnections to considered node X and the sentiment related to a node which is inputted newlineas its weight plays crucial factors. From the experimentation, it is deduced that infection newline(information cascade) propagated by the top N users ranked by the proposed newlineranking method SWPR is high and quicker than traditional metrics