Network Analysis of Digital and Social Media

Permanent URI for this collection


Recent Submissions

Now showing 1 - 5 of 5
  • Item
    Virality and the Virus: Drugs and Cures on Twitter During the COVID-19 Crisis in India
    ( 2022-01-04) Mishra, Dibyendu ; Akbar, Syeda Zainab ; Shekhawat, Gazal ; Pal, Joyojeet
    Social media platforms often become environments of information ambiguity amidst crisis events. We studied the discussion around four "cures" for COVID-19 in India, site of the highest number of recorded positive cases, between 2020 and May 2021, focusing on the role played by high network accounts on social media such as those of journalists, politicians, and celebrities. We find that information paucity led to social media influencers wielding a more important voice online than institutional sources and experts, leading to massive spikes in demand for unproven cures during the peak of the crisis.
  • Item
    Understanding the ego network structure of followers in social marketplaces: Structural capital or liability?
    ( 2022-01-04) Wang, Shan ; Wang, Fang
    In social marketplaces, follower network is one of the most important digital assets that an online seller owns. The existing literature repeatedly reports the strong positive effect of the number of followers on seller performance. However, to date, limited research has been done to understand the ego network structure of a seller’s follower network. The structure of social network may characterize different resources that a seller can leverage to enhance its sales performance. This research studies three network structural properties, including network density, network component and fragmentation, and network centralization, and their impacts on seller performance. A panel data of 1,150 sellers were collected and analyzed. The results show that network density, and centralization are negatively related to seller performance. This suggests that sellers in social marketplaces should avoid highly dense and centralized network when they build and maintain their follower network.
  • Item
    Effects of Random Errors on Graph Convolutional Networks
    ( 2022-01-04) Ando, Shinnosuke ; Tsugawa, Sho
    The use of Graph Convolutional Networks (GCN) has been an emerging trend in the network science research community. While GCN achieves excellent performance in several tasks, there exists an open issue in applying GCN to real-world applications. The issue is the effects of network errors on GCN. Since real-world network data contain several types of noises and errors, GCN is desirable to be less affected by such errors. However, the effects have not been sufficiently evaluated before. In this paper, we analyze the effects of random errors on GCN through extensive experiments. The results show that the node classification accuracy of GCN is decreased only 5% even when 50% of the edges are randomly increased or decreased. Moreover, in terms of false labels, the accuracy of node classification is decreased only 10% even when 20% of the labels are changed.
  • Item
    Applying an Epidemiological Model to Evaluate the Propagation of Toxicity related to COVID-19 on Twitter
    ( 2022-01-04) Maleki, Maryam ; Arani, Mohammad ; Mead, Esther ; Kready, Joseph ; Agarwal, Nitin
    The prevalence of social media has increased the propagation of toxic behavior among users. Toxicity can have detrimental effects on users’ emotion and insight and disrupt beneficial discourse. Evaluating the propagation of toxic content on social networks such as Twitter can provide the opportunity to understand the characteristics of this harmful phenomena. Identifying a mathematical model that can describe the propagation of toxic content on social networks is a valuable approach to this evaluation. In this paper, we utilized the SEIZ (Susceptible, Exposed, Infected, Skeptic) epidemiological model to find a proper mathematical model for the propagation of toxic content related to COVID-19 topics on Twitter. We collected Twitter data based on specific hashtags related to different COVID-19 topics such as Covid, Mask, Vaccine, and Lockdown. The findings demonstrate that the SEIZ model can properly model the propagation of toxicity on a social network with relatively low error. Determining an efficient mathematical model can increase the understanding of the dynamics of the propagation of toxicity on a social network such as Twitter. This understanding can help researchers and policy-makers to develop methods to limit the propagation of toxic content on social networks.
  • Item
    Introduction to the Minitrack on Network Analysis of Digital and Social Media
    ( 2022-01-04) Rosen, Devan ; Chu, Kar-Hai ; Barnett, George