With ambiguous role definitions and without traditional organizational control or coordination mechanisms, how could online community channel knowledge coproduction behaviors among participants with heterogeneity roles? By using a novel theory-building method combining machine learning and narrative analysis, we inves-tigate the heterogeneity of participant roles and knowledge coproduction channeling behaviors in a Chinese self -organizing online community-Miui.com, launched by Xiaomi Inc. Using latent Dirichlet allocation topic modeling analysis and a complementary qualitative analysis of community participants' online forum posts, we initially identify three types of participant role in the community: leaders, supporters, and integrators, and find that (a) conceptualizing narrative practice is adopted to promote knowledge collaboration for leaders; (b) serializing and anthologizing narrative practices are used to promote knowledge collaboration for supporters; and (c) anthologizing narrative practice is adopted to promote knowledge collaboration for integrators. Our study advances the theoretical understanding of the knowledge coproduction and value creation in online communities.
展开▼