Abstract:Virtual communities with its openness and sharing nature, provide a convenient platform for users to participate and share knowledge. Thus users become the core motive for generating content in virtual community. Based on related research on innovation, knowledge sharing and virtual communities, the empirical model considering users’ knowledge sharing level and participation behavior was proposed. Take MIUI community as an example, using the Python web crawlers to get data in "new feature discussion" section from 2011 to 2014, by leveraging Logistic and Poisson regression to perform the analysis. This study reveals the relationship between user participation and knowledge sharing levels, and in which user participation can be measured via users’ contribution value and participation level. Finally use Matlab clustering analysis and divide users into four types according to features of users participation behavior.