SCALE: AI Framework Extends OpenAlex Taxonomy with Fine-Grained Scientific Concepts

A new framework called SCALE (Scientific Concept Aggregation via LLMs and Embeddings) has been developed by researchers to enhance the OpenAlex taxonomy by introducing an additional level of scientific Concepts beneath Topics. This system aims to overcome the shortcomings of current classification methods that fail to adequately represent the nuanced conceptual landscape of modern science. While author keywords provide detail, they often suffer from issues like fragmentation, redundancy, and variability in terminology, rendering them unreliable for consistent knowledge organization. SCALE effectively groups semantically related terms into clear, interpretable units and fits them into the existing disciplinary structure. Utilizing scientific text embeddings, large language models, and graph-based community detection, this framework is poised to significantly enhance the organization of scientific knowledge, thereby aiding research discovery and analysis. The paper can be found on arXiv with the identifier 2608.07254.