Abstract: 【 Purpose/significance】This study takes the domain of natural language processing as a case to investigate the influence of the institutional composition of collaborative teams in international research cooperation among different countries. It aims to guide re⁃ searchers in selecting international partners and enhance their academic impact【. Method/process】The study utilizes paper data from three top-tier conferences in the natural language processing field as the dataset to analyze the impact of institutional composition on academic performance in international research teams. It also compares the differences in institutional composition and academic im⁃ pact among international collaborative teams from the United States, China, and the United Kingdom, which have the highest publica⁃ tion rates.【Result/conclusion】This study finds that in international cooperation, those composed of large enterprises partnering with large enterprises and those composed of a mix of large enterprises, non-large enterprises, high-level research institutions, and normal research institutions tend to yield higher academic influence. In comparison to international collaborative papers with other countries, high-level universities contribute to higher academic influence for international collaborative papers involving the UK, while large en⁃ terprises contribute to higher academic influence for international papers involving the US【. Innovation/limitation】This study innova⁃ tively examines the relationship between institutional composition and academic influence by sub-dividing institution types. Future re⁃ search can further refine institution types and incorporate additional metrics, such as novelty, to comprehensively assess academic im⁃ pact.