Man-machine hybrid response method, system and device

Abstract

The invention belongs to the technical field of man-machine conversation, particularly relates to a man-machine hybrid response method, system and device, and aims to solve the problem that online learning cannot be realized by an existing man-machine response method. The method comprises the following steps of: encoding a current conversation context Ct to obtain a first representation vector E (Ct); Based on a candidate reply statement under the conversation task, and carrying out encoding to obtain a second representation vector E (Ct) , based on the first representation vector E (Ct), wherein the second representation vector E (Ct) is a positive integer greater than or equal to 1. a second representation vector (shown in the specification) obtains the confidence coefficient of the candidate reply statement capable of correctly replying to the question of the user through an uncertainty estimation method, and the confidence coefficient of the candidate reply statement capable of correctly replying to the question of the user is obtained through the uncertainty estimation method; and if the confidence coefficient is greater than a set threshold value, selecting a candidate replystatement corresponding to the confidence coefficient to carry out response output, otherwise, obtaining the reply statement input by the human-computer interaction equipment or the selected candidatereply statement to carry out response output, and carrying out parameter optimization in the above steps based on all dialogue statements obtained after response output. According to the method, it is guaranteed that the output response statement has enough confidence, and online learning and updating of the dialogue model are achieved.

Description

technical field [0001] The invention belongs to the technical field of man-machine dialogue, and in particular relates to a man-machine mixed response method, system and device. Background technique [0002] A task-based dialogue system is a human-computer interaction system that assists users to complete tasks in a specific field through natural language interaction. The first step in designing a data-driven task-based dialogue system is to define the user's needs under the task, and then collect corresponding training data to train the dialogue model according to the defined user needs. [0003] When the dialogue system is deployed online, if there is a deviation between the pre-defined user needs and the needs of real users, the system will give incorrect responses when encountering unconsidered needs. Developers can usually alleviate this problem with a more detailed requirements analysis. However, developers can never guarantee that all user needs are considered. In ...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the above-mentioned problems in the prior art, that is, in order to solve the problem that the existing human-computer response method cannot realize online learning, the first aspect of the present invention proposes a human-computer hybrid response method, including:

Images