PREDICTING THE PULSE OF TWITTER: A HETEROGENEOUS BASS MODEL PERSPECTIVE

Abstract

The rapid growth of online social media has transformed Twitter into a global platform where ideas, opinions, and information spread at unprecedented speeds. Predicting the popularity of tweets is essential for applications in digital marketing, trend forecasting, and information dissemination analysis. This study presents a novel heterogeneous Bass model framework tailored to capture the dynamic and nonlinear diffusion patterns of tweets. Unlike traditional popularity prediction methods, which often rely on static user engagement metrics, the proposed approach incorporates heterogeneity in user influence, retweet behavior, and temporal engagement to more accurately model diffusion. By integrating network characteristics and temporal features with the Bass model, the system enhances predictive accuracy and provides deeper insights into factors driving virality. Experimental evaluations on large-scale Twitter datasets demonstrate that the heterogeneous Bass model consistently outperforms baseline statistical and machine learning methods, offering robust and interpretable predictions. The results highlight the potential of this approach to empower businesses, researchers, and policymakers with actionable insights into online social dynamics.