Abstract: 【Purpose/significance】To explore the correlation between interactive argumentation and group collective evolution in open innovation communities, to expand the applied research of collective intelligence theory, and to provide theoretical support for optimiz⁃ ing the interactive practice of open innovation communities and enhancing innovation performance【. Method/process】Using the 23,029 product discussion texts of "Xiaomi Cloud Service" as a dataset, the BERTopic algorithm and cosine similarity calculation are applied to identify the group cognitive evolution patterns, and the distribution laws and differences of argument distribution, argument combina⁃ tion and argument completeness under different evolution patterns are explored through argument annotation.【Result/conclusion】 Based on the interactive text, 53 collective cognitive points are mined, and five evolutionary modes and four typical cognitive evolution⁃ ary paths are identified: fusion, differentiation, inheritance, newborn, and extinction. Interactive argumentation is closely related to the evolution of collective cognition, and the combination of arguments in different evolutionary modes varies significantly, with the evolu⁃ tionary modes of new birth and differentiation presenting higher argumentation completeness【. Innovation/limitation】Combining argu⁃ mentation analysis and the idea of collective cognitive evolution provides new ideas for understanding the mechanism of information interaction on innovation performance in open innovation. In the future, the interaction data test in other innovation fields will be ap⁃ plied to refine new argumentation evaluation indexes.