Predicting the Popularity of Information on Social Platforms without Underlying Network Structure

Area of Science:

• Social Network Analysis
• Information Diffusion Modeling
• Computational Social Science

Background:

• Predicting information cascade size is crucial for online decision-making and viral marketing.
• Traditional methods struggle with complex, multilingual data or inaccessible network structures.
• Existing approaches often fail to accurately capture the dynamics of information spread across platforms.

Purpose of the Study:

• To develop a novel algorithm for predicting the size of information cascades in online social networks.
• To address limitations of traditional methods in handling diverse online content and network data.
• To provide an accurate and efficient method for forecasting long-term content popularity based on early engagement.

Main Methods:

• Empirical research on data from WeChat and Weibo social networking platforms.
• Development of an activate-decay (AD)-based algorithm leveraging early repost amounts.
• Testing the algorithm's ability to fit propagation trends and predict long-term dynamics.

Main Results:

• The information-cascading process is characterized by an activate-decay dynamic.
• The AD-based algorithm accurately predicts long-term content popularity using only early repost data.
• A strong correlation exists between peak and total information dissemination, improving prediction accuracy.
• The proposed method outperforms existing baseline approaches in popularity prediction.

Conclusions:

• The activate-decay model provides a robust framework for understanding information cascades.
• The AD algorithm offers a practical solution for predicting online content popularity with high accuracy.
• Identifying the peak dissemination point significantly enhances predictive capabilities for information spread.