Abstract: This paper aims to explore the influence mechanism of different types of e-commerce streamers on consumers’ purchase from interview data and bullet-screen comments. Study 1 studied the interview data through grounded theory, and obtained the attributes of e-commerce streamers affecting consumers’ purchase; Study 2 specified the research of different types of e-commerce streamers’ attributes on consumers’ search and purchase of recommended products through Python data mining and econometric model regression. The results of Study 1 show that streamers’ attributes of charm, interaction, recommendation, display, professionalism and trust affect consumers’ purchase intention; E-commerce streamers’ attributes affect consumers’ internal perception and further consumers’ purchase intention. It is found that streamers’ attributes and content consistency positively regulate the impact of e-commerce streamers’ attributes on consumers’ internal perception. The results of Study 2 show that the attributes of expert streamers to enhance consumers’ purchase intention are charm, interaction, recommendation, display, professionalism and trust; The attributes of star streamers to enhance consumers’ purchase intention are charm and interaction; The attributes of official store streamers to enhance consumers’ purchase intention are interaction, display, recommendation and trust.

Key words: e-commerce streamer, grounded theory, fine-grained emotion analysis, econometric model, purchase intention