When SMILES Smiles, Practicality Judgment and Yield Prediction of Chemical Reaction via Deep Chemical Language Processing

Simplified Molecular Input Line Entry System (SMILES) provides a text-based encoding method to describe the structure of chemical species and formulize general reactions. Considering that reactions have been represented in language form, we present symbol only model generally predict yield organic synthesis reaction without considering complex quantum physical modeling or chemistry knowledge. Our is first deep neural network application treats text segments as embedding representation most recent natural processing. Experimental results show our can effectively reactions, which achieves high accuracy 99.76% on practicality judgment Root Mean Square Error (RMSE) around 0.2 for prediction. work shows great potential automatic prediction under conditions further applications path with least cost.