Chain-of-thought reasoning enhancement through multi-objective optimized contrastive learning with negative examples | Synapse