Abstract: ［Purpose/Significance］High-impact users of social media hold the features of unique content capabilities, charismatic personality, potential energy values and efficient network traffic monetizing capabilities. It is crucial to construct highimpact user profile and visualize the typical characteristics of group members so as to expand the radiation of excellent online culture and provide accurate services for platforms, maintain core users, as well as supervise and guide public opinion. ［Design/Methodology］According to the Super IP theory, the high-impact users were analyzed from four aspects, including personality, content, online traffic and trust mechanism. The label system was built upon users’ fundamental attributes, behavior attributes and value attributes in order to construct a conceptual model for the user profile. Taking high-impact users’ data from the platform of Weibo as a sample, K-means clustering algorithm was used to conduct user classification. ［Findings/Conclusion］The experimental results show that the highimpact users of Weibo can be segmented into three groups. The typical portraits of different high-impact users have been obtained based on the prototypical representatives of cluster centers. According to the characteristics of key groups, public opinion supervision, individualized services and social marketing strategies are proposed for the corresponding platforms. ［Originality/Value］The Super IP theory is combined with high-impact users’ characteristics of social media to build portraits for high-impact users, as well as provide new methods and perspectives for the online public opinion supervision and social media platform operation and development.

Keywords: Social media, High-impact user, User profile, Super IP, K-means clustering algorithm