Dialogue control method and system

Abstract

The embodiment of the invention provides a dialogue control method and system. A natural language understanding technology is used to provide dialogue control for a voice interaction system. Accordingto intention understanding and sentiment analysis input by a user, a single-round dialogue script or a multi-round topic script is selected, topic changing, question asking, hobbies and the like arejudged according to interaction occurrence conditions, a complete process is achieved, natural and smooth dialogue is achieved, and psychological characteristics of people are met. And meanwhile, thesystem performs modeling on the obtained data by utilizing big data, and tracks preference topics of the user and the real-time popularity of each topic to perform intelligent topic recommendation.

Description

technical field [0001] The invention relates to the technical field of dialogue control, in particular to a dialogue control method and system. Background technique [0002] At present, there are two common dialogue methods: one is a question-and-answer single-round dialogue, which relies on the text entered by the user, and the background search script gives a reply; the other is to write a multi-round question-and-answer script by predicting the user's possible reply. A multi-round dialogue thus carried out. The application scenarios of the two are not exactly the same, and it seems blunt and unnatural to use the same dialogue method in all situations. A reasonable combination of the two can make the dialogue interaction between the user and the system more natural after integration and unification. [0003] The existing scheme has the following problems: [0004] The single-round dialogue method adopts coping strategies and gives responses to user input. This type of di...

AI Technical Summary

Problems solved by technology

[0004] The single-round dialogue method adopts coping strategies and gives responses to user input. This type of dialogue method does not combine user context and relies on user input. It is a passive dialogue and cannot conduct systematic and in-depth dialogues around specific topics. It is only suitable for inquiries, Instrumental interaction occasions such as broadcasting limit the use in many scenarios

[0005] Multi-round dialogue, relying on multi-round dialogue scripts, predicting the possible input of the user, and organizing reply information according to the possible input of the user, but the user's reply cannot be exhaustive, and user input beyond the multi-round dialogue script will often get inappropriate Reply, only relying on multiple rounds of dialogue will make the dialogue system rigid and lose intelligence

[0006] The combination of single-round and multi-round is currently mostly based on the simultaneous matching of single-round and multi-round for user input, and one of them is selected for reply according to the level of confidence. This method simply combines single-round and multi-round dialogues Added together, not only failed to solve the problem of the naturalness of dialogue interaction, but also have the disadvantages of both

Images