Question and answer model training method, question statement processing method and device and storage medium

Abstract

The invention provides a question and answer model training method, which comprises the steps of determining a hidden variable of a word level corresponding to a question statement through a hidden variable generation network of a question and answer model; decoding the word-level hidden variables through a reply statement generation network of the question and answer model to generate reply wordswith the same number as the word-level hidden variables; generating reply sentences corresponding to the hidden variables of the word level according to the reply words with the same number as the hidden variables of the word level; updating the parameters of the hidden variable generation network through the harmonic average of the precision rate and the recall rate corresponding to the generated reply statement; and updating the parameters of the reply statement generation network through the correct data annotation of the reply statement with the minimum loss value. The invention further provides a question statement processing method and device and a storage medium. The reply generated by the question and answer model obtained through training can be more targeted, so that the richness and the foresight of the reply can be improved, and the use experience of a user is improved.

Description

technical field

[0001] The invention relates to data processing technology, in particular to a question-and-answer model training method, a question statement processing method, a device and a storage medium. Background technique

[0002] Human–Computer Interaction (HCI) refers to the use of a certain dialogue language between humans and computers to determine the information exchange process between humans and computers in a certain interactive way. With the development of the human-computer interaction technology, more and more intelligent products based on the human-computer interaction technology emerge as the times require, such as chatterbots and the like. These smart products can chat and communicate with users, and generate corresponding answer information according to users' questions. However, the generated answer information is usually a high-frequency general answer, and these answers are often meaningless. At the same time, related technologies can only generat...

Images