KMT Reactions

The Chemical Reaction Database (KMT) is a high-resolution map of synthetic organic chemistry, containing approximately 1.44 million reaction records and 1.6 million unique compounds. Unlike traditional chemical lists, this dataset is structured specifically for machine learning and topological analysis. 1. Data Architecture: Reaction SMILES The core of the dataset uses Reaction SMILES notation. Each entry is a single string divided by "greater than" symbols (>), representing a transition in chemical state: Reactants > Reagents/Catalysts/Solvents > Products This linear format allows the data to be parsed into Molecular Graphs, where atoms are nodes and bonds are edges. By comparing the "Before" and "After" graphs, we can calculate the Reaction Delta—the exact structural and electronic changes occurring on the manifold.