FORECASTING SINGLE TWEET POPULARITY USING SOCIAL DIFFUSION WITH HETEROGENEOUS BASS MODELING

Abstract

The virality and popularity of material on social media platforms, such as Twitter, have substantial consequences across several areas, including marketing, public opinion analysis, and information diffusion research. Predicting the popularity of an individual tweet is intrinsically complicated because to its dynamic nature and reliance on many elements such as content, user influence, and network behaviour. This study presents an innovative forecasting method using the Heterogeneous Bass Model (HBM), an adaption of the traditional Bass diffusion model, to anticipate the popularity trajectory of individual tweets. The suggested approach, in contrast to previous models, incorporates diverse variables including tweet information, user interaction patterns, and temporal activity. By integrating these factors into the diffusion process, the model encapsulates both innovation-oriented and imitation-oriented behaviours. Experimental assessment using actual Twitter datasets indicates that the HBM-based approach surpasses baseline machine learning and time series models for prediction accuracy and temporal stability.