Shulou(Shulou.com)11/24 Report--

CTOnews.com July 3 news, Tencent Technology (Shenzhen) Co., Ltd. and West Lake University jointly applied for "sentence generation method, device and computer-readable storage medium" recently announced.

CTOnews.com inquired from the official website of the State intellectual property Office that the application was submitted on December 23, 2021 and announced on June 30, 2023.

According to the patent specification, with the development of artificial intelligence (AI) technology and the emergence of various man-machine dialogue products, such as dialogue robots, people's daily life has been greatly enriched. In the course of research and practice, the inventor of this application found that in the prior art, in the use of all kinds of man-machine dialogue products, the context content of the man-machine dialogue is weak, and the logic of the context content of the man-machine dialogue is poor. there is usually repetitive or contradictory content in the reply of the context content.

For example, in the existing technology, in the process of using all kinds of man-machine dialogue products, there is often a reply of "the weather is very sunny today" in the content of the former man-machine conversation, while the reply to similar questions in the current man-machine conversation is "it is rainy today, so it is not suitable to go out", the contents of the man-machine conversation contradict each other, and the user experience effect of the man-machine dialogue is poor and the applicability is low.

The application embodiment provides a statement generation method, a device and a computer-readable storage medium, which can improve the generation accuracy of the target response statement of the target question and answer statement, improve the coherence of the target response statement, and have high efficiency of sentence generation and strong applicability.

The application embodiment discloses a statement generation method, a device and a computer-readable storage medium, wherein the method comprises:

When the target question and answer statement is obtained, the history statement of the target question and answer statement is obtained, here, the history statement is a plurality of question and answer statements obtained and / or generated before the target question and answer statement is obtained.

The target question answer statement and the history statement are input into the dialogue model, and a plurality of candidate response statements of the target question answer sentence output from the dialogue model are obtained.

Multiple candidate response statements are inputted into the statement coherence classification model, and the coherence scores of each candidate response statement and historical statement are obtained based on the statement coherence classification model.

The target response statement of the target question and answer statement is determined from a plurality of candidate response statements based on the coherence score of each candidate response statement.

The patent specification shows that the application can improve the accuracy of the target response sentence generation of the target question and answer sentence, improve the coherence of the target response sentence, and has high efficiency and strong applicability.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.