Method and system for keeping conversation continuity of conversation system

Abstract

The invention provides a method and a system for keeping conversation continuity of a conversation system. The method comprises the steps of cutting out a related topic set belonging to the same topic as a current conversation from all history conversations in real time according to current conversation input, digging up a current topic keyword from the related topic set, and determining response output according to the current conversation input and the current topic keyword. According to the method and the system for keeping the conversation continuity of the conversation system, the related topic set belonging to the same topic as the current conversation is cut off from the former topics in real time according to a content of the current conversation input; the most typical current topic keyword is dug up from the cut related topic set; the dug keyword contains context information; the content of the current conversation input and the current topic keyword serve as system input; the conversation continuity of the conversation system can be kept; a man-machine conversation achieves a continuous and in-depth effect.

Description

technical field [0001] The invention relates to a method and system for maintaining dialogue continuity in a dialogue system. Background technique [0002] In the man-machine dialogue system, the number of rounds of chatting between the user and the dialogue system is the most intuitive indicator to measure the quality of the dialogue system. In order to increase the number of rounds of chatting between the user and the dialogue system, the continuity of the dialogue system must be guaranteed. On the one hand, under the chat topic discussed by the user, the dialogue system needs to make relevant replies on this topic as much as possible to avoid answering the question; on the other hand: when When the user switches to a new chat topic, the dialogue system also needs to switch to the new chat topic. [0003] In the prior art, the more commonly used method for maintaining the dialogue continuity of the dialogue system is: to obtain the previous N rounds of historical dialogue...

AI Technical Summary

Problems solved by technology

[0004] However, the disadvantages of the existing technology are: on the one hand, the value of N is difficult to determine, if it is too small, for example, N=1, then only the previous round of historical dialogue information can be included, so it is easy to be limited by local information ; If it is too large, it may exceed the scope of the context and introduce irrelevant information; on the other hand, it is very likely to obtain too many keywords from N rounds of historical dialogues, when keywords with low relevance or even irrelevant keywords are introduced It will bring noise to the dialogue system, and the more irrelevant keywords, the greater the noise, which will seriously affect the dialogue quality

Images