How to Find Opinion Leader on the Online Social Network?

TL;DR

This survey addresses finding opinion leaders in online social networks by bridging social science concepts with graph-based, control-theoretic, and sampling methods. It categorizes methods into topology-based centrality, topic-sensitive centrality, control-based centrality, and graph sampling, detailing representative measures and their theoretical underpinnings, as well as practical implementation in a Twitter case study. The work highlights that different definitions of influence yield divergent rankings, emphasizes the benefit of integrating socio-psychological models with graph-signal analysis, and outlines cross-disciplinary directions and challenges for future research. The study's cross-disciplinary perspective aims to unify concepts of influence, improve topic-specific detection, and support applications in information diffusion and public discourse monitoring.

Abstract

Online social networks (OSNs) provide a platform for individuals to share information, exchange ideas, and build social connections beyond in-person interactions. For a specific topic or community, opinion leaders are individuals who have a significant influence on others' opinions. Detecting opinion leaders and modeling influence dynamics is crucial as they play a vital role in shaping public opinion and driving conversations. Existing research have extensively explored various graph-based and psychology-based methods for detecting opinion leaders, but there is a lack of cross-disciplinary consensus between definitions and methods. For example, node centrality in graph theory does not necessarily align with the opinion leader concepts in social psychology. This review paper aims to address this multi-disciplinary research area by introducing and connecting the diverse methodologies for identifying influential nodes. The key novelty is to review connections and cross-compare different multi-disciplinary approaches that have origins in: social theory, graph theory, compressed sensing theory, and control theory. Our first contribution is to develop cross-disciplinary discussion on how they tell a different tale of networked influence. Our second contribution is to propose trans-disciplinary research method on embedding socio-physical influence models into graph signal analysis. We showcase inter- and trans-disciplinary methods through a Twitter case study to compare their performance and elucidate the research progression with relation to psychology theory. We hope the comparative analysis can inspire further research in this cross-disciplinary area.

Figures (2)

• Figure 1: Relationship Among Four Methodologies. The Topology-based Centrality methodologies (3.1) focus on the graphical structure of the network. Real-time content of user $i$ can be represented as an opinion states vector denoted as $x_i$ using Natural Language Processing (NLP). By incorporating Topical Analysis, the Topic-Sensitive Centrality methods (3.2) integrate topical information with the graph structure to identify opinion leaders within specific topics. In opinion evolution modelling, the influence from User $i$ to User $j$ can be modelled as a function $g(x_i,x_j)$. Given the dynamic influence model, the Control methodologies (3.3) aim to find individuals who have the ability to steer the opinion direction by considering the control signals $u$. The Graph Sampling methodologies (3.4) seek to identify specific users who can accurately reconstruct the entire opinion network.
• Figure 2: Opinion Evolution Modeling, interlinking four methods: The Opinion Evolution Modeling formula at the center delineates the mathematical model underpinning opinion dynamics. Topology-based detection emphasizes the maximization of spread on an unbiased dynamic represented solely by graph structure $A_{ij}$. Topic-sensitive detection, where the maximization of spread on a biased dynamic is contingent on the topical relevance of the content, signified by the graph signal $x_i$ based on topic. By constructing opinion evolution models as $\dot x_i = f_i(x_i)+\sum_{j\neq i}^n A_{ij}g(x_i,x_j)$, graph sampling methods identify orthogonal influencers to reduce redundancy in message dissemination across the network. Control theory approaches aim to maximize the size of the controllable subsystem in the network, by incorporating control signal $u_i$.