Mapping the research landscape of artificial intelligence for knowledge discovery in innovation research

Introduction. Artificial intelligence (AI) is increasingly vital for knowledge discovery within innovation research (IR) and the science of science (SoS), yet its specific research landscape lacks systematic mapping. This study addresses this gap by providing a comprehensive bibliometric overview of AI's application for knowledge discovery in innovation research, aiming to structure the field and identify key trends. Method. A bibliometric analysis was performed on 1,094 articles and reviews published between 2010 and 2024, retrieved from the Web of Science Core Collection. Data processing and visualisation employed VOSviewer and Bibliometrix. Analysis. Descriptive statistics quantified publication growth and collaboration patterns. Network analyses mapped thematic structures, using keyword co-occurrence; identified intellectual foundations, through co-citation networks; and visualised current research frontiers through bibliometric coupling. Results. Findings indicate exponential publication growth and high international collaboration, dominated by China and the USA. Key thematic clusters focus on AI methodologies (machine learning (ML), deep learning (DL), natural language processing (NLP)), innovation contexts (patent analysis, technology trends), and integrated science of science methods (bibliometrics, scientometrics). Intellectual foundations derive strongly from computer vision, sequence/topic modelling, and bibliometric tools. Conclusion. This mapping structures the field, highlighting AI's profound integration as both a transformative tool for innovation analysis and an object of study within the science of science framework itself. It underscores the field's dynamism and provides a basis for future research on AI's impact and responsible application.