Stack-based chat robot dialogue management method

Abstract

The invention discloses a stack-based chat robot dialogue management method, and relates to the technical field of multi-round dialogues. The method comprises the following steps: S1, a user session enters a stack: a user is allocated with an identification sessionId, and the sessionId is pressed into a session stack; S2, a multi-round dialogue stack is operated: an intention of the user is identified and the identified intention is pressed into a multi-round dialogue; and S3, a word slot filling stack is operated: slot information of the intention is filled after the intention of the user isdetermined, and until all slots are filled, an execution action of the intention is triggered. According to the method, the user session is put into the stack in robot dialogue management; the intention of the user is identified by a context cache and pressed into the multi-round conversation stack; the slot information of the intention is filled according to the intention of the user; the next slot needing to be filled can be popped out from the word groove filling stack after each slot filling is successful; and until all the slots are filled, the execution action of the intention is triggered, so that the dialogue processing efficiency and the user communication capacity of a robot are improved.

Description

technical field [0001] The invention belongs to the technical field of multi-round dialogues, in particular to a stack-based chat robot dialogue management method. Background technique [0002] Today, ASR and TTS technologies are relatively mature, and the expression and understanding of natural language has made great progress, and it has gradually weakened in the industry's competitive barriers. The core competitiveness of future intelligent dialogue robots lies in understanding the user's intentions For the differentiated services provided later, a comprehensive dialogue management solution plays a key role in the above-mentioned services. [0003] The intelligent dialogue system can be roughly divided into five basic modules: Speech Recognition (ASR), Natural Speech Understanding (NLU), Dialogue Management (DM), Natural Language Generation (NLG), and Speech Synthesis (TTS), so the dialogue management module is a Very important module. Speech recognition converts speech...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to provide a stack-based chat robot dialog management method, by pushing the user session into the stack in the robot dialog management, using the context cache to identify the user's intention and pressing it into the multi-round dialog stack and filling the word slots to solve the problem Existing robots require users to fully state their requirements each time before they can be analyzed, resulting in low analysis and processing efficiency and low intelligence

Images