A dictionary-based approach to efficiently compress SMILES-based datasets, enabling random access and readable output.
