In order to understand the knowledge of Undergraduates in Chengdu about electronic cigarette, strengthen the health propaganda and education of electronic cigarette, prevent and control electronic cigarette, and build a smokeless campus environment, this paper aims to analyze the recognition of Undergraduates in Chengdu on the reliability of electronic cigarette sources based on data mining technology. Firstly, using random sampling method, a university in Chengdu was selected as the research object, and the attempted smoking rate, electronic smoke awareness rate and electronic smoke usage rate of college students were analyzed. The main factors that affect college students' awareness of electronic smoke were analyzed, and the reliability of electronic smoke sources was analyzed. Then, based on online review text mining method, the confidence of electronic smoke sources was analyzed. This paper builds a theme model to analyze college students' perception of the reliability of e-cigarette sources by using classification topics. Through experimental validation, the method proposed in this paper has higher data mining accuracy and efficiency, can effectively improve college students' awareness of electronic cigarette, strengthen the health promotion of electronic cigarette, and expect to provide support for subsequent related research.