Purpose Facing the increasingly complex realities and scientific challenges, the demand for interdisciplinary collaboration is growing. However, merely assembling researchers from diverse disciplinary backgrounds does not guarantee innovative outcomes that meet expectations. This study aims to analyze the scientific cooperation networks in the field of natural language processing and identify interdisciplinary collaboration structures with innovative driving forces. Design/methodology/approach This study constructed a scientific collaboration network based on 14,646 publications from eight leading enterprises in the field of natural language processing and conducted the following analysis: Firstly, the disciplinary attributes of scientists are determined by using the method of disciplinary contribution. Secondly, the disciplinary attributes of scientists are encoded with different colors, and colored motif analysis is used to identify key collaborative patterns at the mesoscopic level of networks. Finally, the impact of colored motif frequency on disruptive innovation is analyzed in the team subnetwork. Findings These analyses indicate that (1) in Natural Language Processing, more than 70% of scientists have a background in computer science. (2) Across the 2010–2023 collaboration network, we identify 11 significant three-node motifs, six of which are interdisciplinary. (3) Notably, the “Multidisciplinary–Computer science–Computer science (CM3)” collaboration pattern exhibits substantial potential for fostering innovation. (4) Furthermore, under the influence of this collaboration pattern, certain differences in innovation performance can be observed between Chinese and U.S. enterprises. Originality/value Most of the existing interdisciplinary analysis studies are measured by disciplinary diversity indicators, thus lacking more microscopic discussions. Network motif can discuss specific interdisciplinary collaboration structures from a mesoscopic network perspective. The collaboration pattern of “based on mainstream disciplines and efficiently linking and integrating knowledge from multiple disciplines” discovered in this study can not only enrich the theory of interdisciplinary collaboration but also provide practical strategies for the formulation of research policies and team building.
You et al. (Fri,) studied this question.