Added Objective Guided Optimization of Adversarial Text Generation Method

The problem of error accumulation is caused by the supervision deep neural network text generation model. In 
 order to solve this problem, a model based on reinforcement antagonistic thought training proposed.The adversarial can be generated proposed model, and then used for identification, learning reward function optimized, optimized reduce probability accumulation.More structure knowledge added into integrating target guidance feature actual process make have higher authenticity. paper, author optimizes method basis target-guided optimization, which reference practitioners.