The rapid growth of live streaming has intensified competition among streamers. A key challenge lies in aligning live-streaming sales strategies with consumer needs to cultivate positive consumer attitudes.
Drawing on theories of the types of consumer–live streamer interaction strategies, the uses and satisfaction theory, the channel complementarity theory and the congruence hypothesis of live streaming content and online reviews, this study employs quantitative research to investigate how to effectively integrate genuine product feature preferences, as reflected in online reviews, with key messages conveyed by live streamers. To achieve the objective, this paper synthesizes machine learning techniques – including information quantity calculation and BERT – with statistical analysis methods such as difference testing and scenario experiments.
The results demonstrate that the proposed predictive model of perceived useful information, which relies on information quantity, provides notable advantages. Specifically, there is a significant disparity between consumer feature preferences derived from useful information and those identified through traditional methods that consider all available data. Moreover, feature preferences extracted from useful information diverge from the focal points of the live streamer’s actual explanations. The proposed method, which aligns consumer preferences with the live streamer’s explanations, markedly enhances consumers’ purchase intentions.
This study introduces a novel metric of information quantity to distinguish useful reviews from the broader dataset, integrating it with sentiment analysis and feature extraction techniques. Unlike traditional methods that treat all reviews as equally valuable, this approach prioritizes comments based on informational value, emotional tone and relevance to product features, providing a nuanced and precise understanding of consumer preferences. By aligning these preferences with live streamers’ explanations, the study offers a data-driven, consumer-centric strategy for improving product recommendations and engagement in live commerce. This comprehensive framework represents a significant advancement over traditional models reliant on data volume alone.
