Yaozong Li, Weina Lin, Yuanyue Feng# (# corresponding author)
Accepted by the Doctoral Consortium of the 24th Wuhan International Conference on E-Business (WHICEB) 2025-05-11
[Purpose/Significance] This study aims to deepen the understanding of how real-time comments influence sales performance in the context of live streaming e-commerce. [Methodology/Process] A theoretical framework was constructed based on the Elaboration Likelihood Model (ELM). Utilizing real data from 52 livestreams on the Douyin platform, two deep learning-based text classification models were trained and employed to conduct an in-depth analysis of comments. A generalized linear model (GLM) was then employed for empirical analysis. [Findings/Conclusion] It was found that comment length has a significantly negative effect on sales, whereas sentiments and product attention exhibit an inverted U-shaped relationship with sales, and the number of comments has a significantly positive impact on sales. The model demonstrated high explanatory power (Pseudo R² = 0.8249). [Originality/Value] Cutting-edge natural language processing (NLP) methods were applied to empirical studies in live streaming e-commerce, fostering interdisciplinary integration and advancing methodological innovation in the field. The conclusions, grounded in real-world data, offer valuable insights and provide a theoretical foundation for merchants to optimize their livestreaming strategies.
Yaozong Li, Weina Lin, Yuanyue Feng# (# corresponding author)
Accepted by the Doctoral Consortium of the 24th Wuhan International Conference on E-Business (WHICEB) 2025-05-11
[Purpose/Significance] This study aims to deepen the understanding of how real-time comments influence sales performance in the context of live streaming e-commerce. [Methodology/Process] A theoretical framework was constructed based on the Elaboration Likelihood Model (ELM). Utilizing real data from 52 livestreams on the Douyin platform, two deep learning-based text classification models were trained and employed to conduct an in-depth analysis of comments. A generalized linear model (GLM) was then employed for empirical analysis. [Findings/Conclusion] It was found that comment length has a significantly negative effect on sales, whereas sentiments and product attention exhibit an inverted U-shaped relationship with sales, and the number of comments has a significantly positive impact on sales. The model demonstrated high explanatory power (Pseudo R² = 0.8249). [Originality/Value] Cutting-edge natural language processing (NLP) methods were applied to empirical studies in live streaming e-commerce, fostering interdisciplinary integration and advancing methodological innovation in the field. The conclusions, grounded in real-world data, offer valuable insights and provide a theoretical foundation for merchants to optimize their livestreaming strategies.