Abstract
Live-streaming commerce has fundamentally reshaped online purchasing behavior by merging real-time interaction with digital consumption.This study advances the classical Stimulus-Organism-Response (S-O-R) framework by embedding artificial intelligence-driven emotion recognition to quantify consumersꞌ affective and social responses.Drawing on data from 1443 Douyin users engaged with Hongqi Chainꞌs livestream sessions, we employed logistic regression and robustness validation in R to examine how streamer interactivity, professionalism, and attractiveness affect emotional response and social presence -two key organism constructs -there by influencing purchase intention.The inclusion of AI-derived emotion metrics, obtained from a fine-tuned Chinese BERT sentiment model, improved the modelꞌs explanatory power by 5.6% and raised the adjusted R² to 0.861.Social presence emerged as the dominant mediator linking streamer stimuli to behavioral intention.The findings reconceptualize the organism stage of the S-O-R model within an AI-enhanced analytical paradigm and provide managerial implications for optimizing live-streaming strategies through datadriven consumer insight.