Knowledge-driven easy-to-configure task-oriented dialogue management method and system

Abstract

The invention relates to a knowledge-driven easy-to-configure task-oriented dialogue management method and system. The system mainly comprises the following contents: 1, a configurable dialogue strategy tree module, a dialogue strategy tree structure is designed, system nodes and function nodes are configured, and the output action of the system is determined according to the current dialogue state and a dialogue strategy tree during dialogue strategy selection; and 2, an FAQ knowledge base and query module: for the user asking questions not in the action space defined by the system, the replyof the system comprises two parts, one part is the reply obtained through knowledge base query, and the other part is the next step of system response selected according to the dialogue strategy. Knowledge base query is carried out by a hierarchical query method. Firstly, semantic matching is carried out on user questions and candidate questions, and then accurate reply is carried out according to fine-grained information in a dialogue state. Rapid migration and configuration of a dialogue scene and task-oriented dialogue control are realized through introduction of two pieces of external knowledge.

Description

technical field [0001] The invention relates to the technical field of human-machine natural language dialogue, and is used to construct a machine automatic customer service application system, specifically a knowledge-driven, easy-to-configure, task-oriented dialogue management method and system. Background technique [0002] With the development of artificial intelligence technology and natural language processing technology, dialogue systems are more and more widely used in industries such as intelligent customer service. According to the "China Smart Customer Service Industry Research Report" data, the current stock market size of China's customer service software is between 10 billion and 15 billion yuan, and smart customer service is replacing manual customer service at a rate of 40%-50%. At present, there are two main types of dialogue systems: chatting dialogue systems and task-oriented dialogue systems. [0003] The task-oriented multi-round dialogue system aims to...

AI Technical Summary

Problems solved by technology

[0005] However, the current dialogue management module still has many disadvantages: first, it is difficult to migrate, when switching the dialogue system to a new scene, the dialogue strategy needs to be redesigned; second, it is difficult to maintain, for the same dialogue scene, changing the dialogue strategy consumes Time-consuming and labor-intensive; third, the user experience is poor. Since the action space defined by the system is limited, when the user asks an unrecognized question, in order to control the task-oriented dialogue, the system often ignores the question or gives an ambiguous reply

The user's sense of participation and experience is not enough, and the satisfaction is reduced

Images