Processed Data for Semantic Mapping

Processed data and result files for a bibliometric mapping study of the scientific literature on recycling. This version adds the results of a supervised fine-tuning experiment (SBERT / all-MiniLM-L6-v2) evaluated on cross-cluster citation retrieval: main experiment, a three-configuration robustness check, and an 80/20 dissociation split (held-out intra-cluster vs. cross-cluster recall). See finetuning_files.zip and its README.md for the file-level description, metrics, and training-set sizes. Code to reproduce the results is available at https://github.com/anabossler/bibliometric_api.