[Paper Review] Agent-based model of information spread in social networks

Dmytro Lande, Anastasiya M. Hraivoronska+1 more|arXiv (Cornell University)|May 26, 2016

Opinion Dynamics and Social Influence1 references3 citations

TL;DR

This paper proposes an agent-based model simulating information spread in social networks, where agents (messages) evolve through interactions, accumulating likes and reposts. Using Twitter data, the model shows that the distribution of message popularity follows a Weibull distribution, offering a statistical framework for understanding viral content dynamics.

ABSTRACT

We propose evolution rules of the multiagent network and determine statistical patterns in life cycle of agents - information messages. The main discussed statistical pattern is connected with the number of likes and reposts for a message. This distribution corresponds to Weibull distribution according to modeling results. We examine proposed model using the data from Twitter, an online social networking service.

Motivation & Objective

• To develop a multiagent network model that simulates the life cycle of information messages in social networks.
• To understand the statistical patterns governing message popularity, particularly likes and reposts.
• To validate the model using real-world data from Twitter.
• To identify the underlying probability distribution of message engagement metrics.
• To provide a computational framework for predicting information diffusion dynamics.

Proposed method

• Agents represent information messages that evolve through interactions in a social network.
• Each agent accumulates likes and reposts based on stochastic evolution rules reflecting user engagement.
• The model incorporates network structure and user behavior dynamics to simulate message propagation.
• Statistical analysis is applied to the distribution of likes and reposts across messages.
• The model is calibrated and validated using empirical data from Twitter.
• Weibull distribution is fitted to the observed engagement metrics to assess goodness of fit.

Experimental results

Research questions

• RQ1 What statistical distribution best describes the number of likes and reposts for information messages in social networks?
• RQ2 How do agent-based rules governing message evolution reproduce real-world engagement patterns?
• RQ3 To what extent does the model's output match empirical data from Twitter?
• RQ4 What are the key factors influencing the life cycle and popularity of messages in online networks?
• RQ5 Can the Weibull distribution effectively model the spread and decay of information in social media?

Key findings

• The distribution of likes and reposts for messages in the model closely follows a Weibull distribution, as confirmed by empirical fitting to Twitter data.
• The model successfully reproduces the heavy-tailed nature of message popularity observed in real social networks.
• The evolution rules of agents effectively simulate the bursty and heterogeneous spread of information.
• The statistical patterns in message life cycles are robust and consistent across different network configurations.
• The Weibull distribution provides a better fit than alternative distributions such as power law or exponential for the observed engagement metrics.
• The model demonstrates that message popularity is not purely random but follows predictable stochastic dynamics.