This study classifies posts into four distinct topics and uses their sentiment values to predict product return rate at e-commerce websites. The results reveal that the sentiments of posts related to e-commerce company news (i.e., objective posts) are negatively related to product return rate. On the contrary, the sentiments of social network posts related to product use, purchase and service experiences (i.e., subjective posts) are positively related to product return rate at a focal e-commerce website. The paper contributes to product return research as well as social network prediction research. Practitioners may use the method to predict product return rates using social network posts.
Ding, Yi; Xu, Haifeng; and Tan, Bernard C. Y., "PREDICTING PRODUCT RETURN RATE WITH “TWEETS”" (2016). PACIS 2016 Proceedings. 345.