Man-machine conversation reply generation method and device

Abstract

The invention provides a man-machine conversation reply generation method and a man-machine conversation reply generation device. The method comprises the steps of acquiring a user message in a man-machine conversation, generating a plurality of keywords for replying the user message, encoding the plurality of keywords by use of an encoder to obtain a first hidden state of the plurality of keywords; sequentially generating a plurality of second hidden states corresponding to text positions of the reply message by use of a decoder according to the first hidden state of the plurality of keywords; sampling according to each of the second hidden states to obtain vocabularies located on the text positions corresponding to the second hidden states; and generating a reply message according to thetext position where each vocabulary is located. According to the man-machine conversation reply generation method and the man-machine conversation reply generation device, the plurality of keywords are generated by acquisition of the user message, the keywords are encoded and decoded, the text vocabularies corresponding to the keywords are acquired through sampling to further generate the reply message, and the problems in the prior art that a specific template is depended on, the generated reply message is poor in applicability, correlation and diversity and the user demand cannot be satisfied are solved.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence, in particular to a reply generation method and device for man-machine dialogue. Background technique [0002] The development of natural language processing technology has enabled effective communication between humans and computers using natural language through man-machine dialogue systems, and reply generation is an important part of man-machine dialogue systems. In man-machine dialogue systems, the system is based on User's message, generate a sentence of natural language as the system reply. [0003] In related technologies, the reply generation technology in the human-computer dialogue system mainly adopts the template method, and there is also an end-to-end generation technology based on deep learning, which generates a corresponding natural language reply according to the received user's message, but the two methods The applicability of the system is poor, and the...

AI Technical Summary

Problems solved by technology

[0003] In related technologies, the reply generation technology in the human-computer dialogue system mainly adopts the template method, and there is also an end-to-end generation technology based on deep learning, which generates a corresponding natural language reply according to the received user's message, but the two methods The applicability of the system is poor, and the fluency, relevance and diversity of the reply language obtained are poor, which cannot meet the needs of users

Images