Method and device for asking rhetorical question for question of user based on dialogue system

Abstract

The embodiment of the invention provides a method and device for asking a rhetorical question for a question of a user based on a dialogue system. The dialogue system comprises a reinforcement learning model. The method comprises the steps: obtaining a first dialogue round based on the dialogue system, wherein the first dialogue round comprises T rounds of dialogues; obtaining T states corresponding to the first dialogue round; obtaining a plurality of feedback score vectors respectively corresponding to a plurality of anti-questions included in the T rounds of dialogues, wherein a plurality of dimensions included in the feedback score vectors respectively correspond to a plurality of predetermined features; and based on the T states, the plurality of questions and the plurality of feedback score vectors, training the reinforcement learning model so as to propose a rhetorical question to the question of the user based on the reinforcement learning model.

Description

technical field [0001] The embodiment of this specification relates to the technical field of dialog systems, and more specifically, to a method and device for asking rhetorical questions to user questions based on the dialog system. Background technique [0002] Recently, people have paid more and more attention to how to use machine learning to better build dialogue systems. The dialogue system realizes the communication between human and machine through computer algorithms, including three types: question and answer type, task type and chat type. For the above-mentioned types of dialogue systems, the user usually asks questions and the dialogue system replies. In some cases, the questions raised by users are often vague, and it is difficult for the dialogue system to find matching standard questions directly from the query database and reply based on the answers to the standard questions. In this case, the dialog system needs to ask the user rhetorical questions to clar...

AI Technical Summary

Problems solved by technology

However, when the reinforcement learning model is used in natural language processing, the feedback is usually sparse, the amount of data used to train the model is not enough, and the initial predictive ability of the reinforcement learning model is poor, so that when the reinforcement learning model is started, the Poor user experience affects business development

Images