arXiv subjects

Bailu Jin

Publications and source records attributed to Bailu Jin.

5 recordsLinked to original sources

arXiv2024

Synthetic Social Media Influence Experimentation via an Agentic Reinforcement Learning Large Language Model Bot

Understanding the dynamics of public opinion evolution on online social platforms is crucial for understanding influence mechanisms and the provenance of information. Traditional influence analysis is typically divided into qualitative assessments of personal attributes (e.g., psychology of influence) and quantitative evaluations of influence power mechanisms (e.g., social network analysis). One challenge faced by researchers is the ethics of real-world experimentation and the lack of social influence data. In this study, we provide a novel simulated environment that combines agentic intelligence with Large Language Models (LLMs) to test topic-specific influence mechanisms ethically. Our framework contains agents that generate posts, form opinions on specific topics, and socially follow/unfollow each other based on the outcome of discussions. This simulation allows researchers to observe the evolution of how opinions form and how influence leaders emerge. Using our own framework, we design an opinion leader that utilizes Reinforcement Learning (RL) to adapt its linguistic interaction with the community to maximize its influence and followers over time. Our current findings reveal that constraining the action space and incorporating self-observation are key factors for achieving stable and consistent opinion leader generation for topic-specific influence. This demonstrates the simulation framework's capacity to create agents that can adapt to complex and unpredictable social dynamics. The work is important in an age of increasing online influence on social attitudes and emerging technologies.

cs.SI↗

arXiv2024

Minimize Control Inputs for Strong Structural Controllability Using Reinforcement Learning with Graph Neural Network

Strong structural controllability (SSC) guarantees networked system with linear-invariant dynamics controllable for all numerical realizations of parameters. Current research has established algebraic and graph-theoretic conditions of SSC for zero/nonzero or zero/nonzero/arbitrary structure. One relevant practical problem is how to fully control the system with the minimal number of input signals and identify which nodes must be imposed signals. Previous work shows that this optimization problem is NP-hard and it is difficult to find the solution. To solve this problem, we formulate the graph coloring process as a Markov decision process (MDP) according to the graph-theoretical condition of SSC for both zero/nonzero and zero/nonzero/arbitrary structure. We use Actor-critic method with Directed graph neural network which represents the color information of graph to optimize MDP. Our method is validated in a social influence network with real data and different complex network models. We find that the number of input nodes is determined by the average degree of the network and the input nodes tend to select nodes with low in-degree and avoid high-degree nodes.

cs.LG↗

arXiv2023

How to Find Opinion Leader on the Online Social Network?

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.

cs.SI↗

arXiv2022

Data Driven Modeling Social Media Influence using Differential Equations

Individuals modify their opinions towards a topic based on their social interactions. Opinion evolution models conceptualize the change of opinion as a uni-dimensional continuum, and the effect of influence is built by the group size, the network structures, or the relations among opinions within the group. However, how to model the personal opinion evolution process under the effect of the online social influence as a function remains unclear. Here, we show that the uni-dimensional continuous user opinions can be represented by compressed high-dimensional word embeddings, and its evolution can be accurately modelled by an ordinary differential equation (ODE) that reflects the social network influencer interactions. Our three major contributions are: (1) introduce a data-driven pipeline representing the personal evolution of opinions with a time kernel, (2) based on previous psychology models, we model the opinion evolution process as a function of online social influence using an ordinary differential equation, and (3) applied Our opinion evolution model to the real-time Twitter data. We perform our analysis on 87 active users with corresponding influencers on the COVID-19 topic from 2020 to 2022. The regression results demonstrate that 99% of the variation in the quantified opinions can be explained by the way we model the connected opinions from their influencers. Our research on the COVID-19 topic and for the account analysed shows that social media users primarily shift their opinion based on influencers they follow (e.g., model explains for 99% variation) and self-evolution of opinion over a long time scale is limited.

cs.SI↗

arXiv2022

Revealing the Excitation Causality between Climate and Political Violence via a Neural Forward-Intensity Poisson Process

The causal mechanism between climate and political violence is fraught with complex mechanisms. Current quantitative causal models rely on one or more assumptions: (1) the climate drivers persistently generate conflict, (2) the causal mechanisms have a linear relationship with the conflict generation parameter, and/or (3) there is sufficient data to inform the prior distribution. Yet, we know conflict drivers often excite a social transformation process which leads to violence (e.g., drought forces agricultural producers to join urban militia), but further climate effects do not necessarily contribute to further violence. Therefore, not only is this bifurcation relationship highly non-linear, there is also often a lack of data to support prior assumptions for high resolution modeling. Here, we aim to overcome the aforementioned causal modeling challenges by proposing a neural forward-intensity Poisson process (NFIPP) model. The NFIPP is designed to capture the potential non-linear causal mechanism in climate induced political violence, whilst being robust to sparse and timing-uncertain data. Our results span 20 recent years and reveal an excitation-based causal link between extreme climate events and political violence across diverse countries. Our climate-induced conflict model results are cross-validated against qualitative climate vulnerability indices. Furthermore, we label historical events that either improve or reduce our predictability gain, demonstrating the importance of domain expertise in informing interpretation.

cs.LG↗