Conversation method and device, electronic equipment and readable storage medium

Abstract

The invention relates to the technical field of artificial intelligence, and discloses a dialogue method and device, electronic equipment and a readable storage medium. The dialogue method comprises the steps: obtaining a target question input by a user; based on a trained reinforcement learning model, determining a feedback action corresponding to the target question, wherein the reinforcement learning model is obtained by training based on at least two rounds of sample conversations and a reward function determined by at least two rounds of sample conversations, and each round of sample dialogue comprises a sample question sentence and a corresponding sample answer sentence; determining a target answer sentence corresponding to the feedback action, and outputting the target answer sentence. According to the dialogue method provided by the invention, the prediction accuracy of the finally trained reinforcement learning model is higher.

Description

technical field [0001] The present application relates to the technical field of artificial intelligence, specifically, the present application relates to a dialog method, device, electronic equipment and readable storage medium. Background technique [0002] With the development of information technology, Internet technology develops accordingly. Users often need to query various information through the Internet to obtain corresponding answers. Therefore, task-oriented dialogue has become increasingly popular, which can be used in telephone customer service, mobile customer service, Among the mobile assistants, basic task-based operations such as booking air tickets and hotel reservations can be completed, greatly reducing the use of human resources. [0003] In task-oriented dialogue, the answer sentence is mainly predicted through the reinforcement learning model, and the reinforcement learning model needs to be trained according to the reward function. At present, the r...

AI Technical Summary

Problems solved by technology

At present, the reinforcement learning model is usually trained according to the results of the dialogue, but the obtained reinforcement learning model cannot accurately describe which steps are correct and which are wrong actions in each round of the process, resulting in the prediction of the reinforcement learning model after training. less accurate

Images