The invention relates to a mobile customer service conversation management method based on state tracking and strategy guidance, and belongs to the field of processing of partially observable markov models and neural networks to natural languages. According to the method, a POMDP model is constructed, a strategy optimization algorithm is guided in combination with design of a problem of deep Q network, real-time tracking of a dialogue state is realized, a next dialogue result is predicted in time, a predicted dialogue strategy is given, an optimal dialogue strategy method is obtained accordingto the strategy optimization algorithm, and an optimal dialogue prediction result is deduced. According to the method, a large amount of labor cost is saved, the working efficiency of customer service conversation is improved, and meanwhile, more convenient service experience is provided for users, and the service quality is improved.