Dynamics of Information Diffusion and Social Sensing

Statistical inference using social sensors is an area that has witnessed remarkable progress and is relevant in applications including localizing events for targeted advertising, marketing, localization of natural disasters and predicting sentiment of investors in financial markets. This chapter presents a tutorial description of four important aspects of sensing-based information diffusion in social networks from a communications/signal processing perspective. First, diffusion models for information exchange in large scale social networks together with social sensing via social media networks such as Twitter is considered. Second, Bayesian social learning models and risk averse social learning is considered with applications in finance and online reputation systems. Third, the principle of revealed preferences arising in micro-economics theory is used to parse datasets to determine if social sensors are utility maximizers and then determine their utility functions. Finally, the interaction of social sensors with YouTube channel owners is studied using time series analysis methods. All four topics are explained in the context of actual experimental datasets from health networks, social media and psychological experiments. Also, algorithms are given that exploit the above models to infer underlying events based on social sensing. The overview, insights, models and algorithms presented in this chapter stem from recent developments in network science, economics and signal processing. At a deeper level, this chapter considers mean field dynamics of networks, risk averse Bayesian social learning filtering and quickest change detection, data incest in decision making over a directed acyclic graph of social sensors, inverse optimization problems for utility function estimation (revealed preferences) and statistical modeling of interacting social sensors in YouTube social networks.

💡 Research Summary

This chapter provides a comprehensive tutorial on four interrelated aspects of information diffusion and social sensing in modern online networks, viewed through the lenses of signal processing, economics, and network science.

1. Network Diffusion Models – The authors adopt the classic Susceptible‑Infected‑Susceptible (SIS) framework to describe how ideas, rumors, or health signals spread among a massive population of social sensors. By applying mean‑field approximations to random graph topologies, the high‑dimensional stochastic dynamics are reduced to tractable ordinary differential equations that capture the evolution of the overall infection (or adoption) rate. Real‑world validation is performed using CDC influenza‑like illness (ILI) data and Twitter hashtag streams, demonstrating that Twitter can serve as a low‑cost, real‑time sensor for epidemic monitoring. The section also reviews two sampling strategies—social sampling and respondent‑driven sampling—highlighting their bias‑reduction properties and their adoption by public‑health agencies.
2. Bayesian Social Learning – Here each agent updates a Bayesian belief about an underlying state by fusing its private noisy observation with the actions (recommendations, ratings) of its predecessors on a directed acyclic graph. The authors extend the standard model to incorporate risk‑averse decision makers using Conditional Value‑at‑Risk (CVaR) as the utility criterion, and they derive a quickest‑change‑detection rule that is robust to market shocks. A novel contribution is the analysis of “data incest,” the phenomenon where the same piece of information circulates through multiple paths, creating spurious correlations that can lead to herding. Sufficient graph‑structural conditions for incest mitigation are presented, and human‑subject experiments illustrate the detrimental impact of incest on collective accuracy.
3. Revealed Preferences and Utility Estimation – Leveraging Afriat’s theorem, the chapter shows how to test whether a finite time‑series of price‑action pairs could have been generated by a utility‑maximizing agent. The test is both necessary and sufficient, and it yields a constructive method for recovering a consistent utility function. The authors extend the result to multi‑agent settings, providing a test for whether observed behavior corresponds to a Nash equilibrium of a concave potential game. Three empirical datasets—an undergraduate auction, Ontario electricity consumption, and Twitter hashtag responses—are examined, confirming that the revealed‑preference framework can uncover underlying utility structures even in noisy social data.
4. YouTube Social Interaction – The final section treats YouTube as a hybrid social‑media/network platform where interaction follows a user‑content‑user pattern. Using over six million videos from 25 000 channels, the authors employ Granger causality, VAR modeling, and time‑series analysis to quantify how meta‑level features (title, thumbnail contrast, tags) and subscriber dynamics drive view counts and engagement. Key findings include: (i) first‑day view count, subscriber growth, and thumbnail contrast are the strongest predictors of long‑term popularity; (ii) changes in meta‑data precede spikes in subscriber activity, indicating a causal direction; (iii) external social‑media events can be modeled as exogenous shocks affecting engagement trajectories. The analysis suggests practical strategies for content creators: dynamically adjust meta‑data, schedule uploads strategically, and leverage cross‑platform promotion to maximize engagement and revenue.
   Overall, the chapter weaves together macroscopic diffusion dynamics, microscopic Bayesian learning, utility‑based revealed preference tests, and concrete YouTube case studies into a unified framework for social sensing. It demonstrates how low‑resolution, quantized decisions from human users can be interpreted, modeled, and ultimately exploited for real‑time event detection, risk‑aware decision support, and optimized content dissemination in today’s interconnected digital ecosystems.