Scientific assessments of current knowledge about the Earth System and its components have become essential for planning, implementing, and evaluating environmental policies such as the Global Biodiversity Framework. With the increasing availability and accessibility of mountain biodiversity data and knowledge worldwide, assessments like those carried out by IPCC or IPBES have become possible to fill an urgent gap in our understanding of the state of and trends in mountain biodiversity research, species, and ecosystems. Here and as part of a first-of-its-kind global assessment of mountain biodiversity, we present a novel approach for systematically mapping scientific literature on mountain biodiversity. Our method integrates text mining, pattern recognition, heuristic algorithms, and an expert-validated vocabulary to enrich publications metadata with spatial, taxonomic, and topical information. This approach enables the creation of a richly annotated database, serving as a foundation for both evidence mapping and a global assessment of mountain biodiversity. While the current workflow remains labour-intensive, requiring expert validation to ensure accuracy, it also lays the groundwork for future AI-driven literature exploration. Specifically, our validated dataset will be used to train machine learning models within the MoBiKo project – a Knowledge Graph for Mountain Biodiversity – enhancing automated knowledge extraction and synthesis. By developing this structured, high-quality knowledge base, we aim to support scientific research, policy and conservation efforts, advocacy, as well as investment for mountain biodiversity worldwide.