公开(公告)号：US20210027177A1

公开(公告)日：2021-01-28

申请号：US16818927

申请日：2020-03-13

Applicant: Advanced New Technologies Co., Ltd.

Inventor： Kaisheng Yao ,  Jiaxing Zhang ,  Jia Liu ,  Xiaolong Li

IPC: G06N5/04 ,  G06F7/58

Abstract: Implementations of the present specification provide a method and an apparatus for extending question and answer samples. According to the method, a random number is generated for each existing sample, a question is blurred for a sample whose random number belongs to sample extension random numbers, to generate an extended sample, so that an overall sample blurring extension rate can be effectively controlled. In addition, for a sample needing blurring extension, a question is extended by deleting a word with a predetermined part of speech in the corresponding question, and then an extended sample is generated based on an extended question, so that more question expression ways are compatible. As such, a question and answer model is trained by using a sample set to which extended samples are added, so that an answer can be provided to a user more effectively.

公开(公告)号：US11100412B2

公开(公告)日：2021-08-24

申请号：US16818927

申请日：2020-03-13

Applicant: Advanced New Technologies Co., Ltd.

Inventor： Kaisheng Yao ,  Jiaxing Zhang ,  Jia Liu ,  Xiaolong Li

IPC: G06F16/90 ,  G06F15/17 ,  G06F7/00 ,  G06Q50/24 ,  G06N5/04 ,  G06F7/58

Abstract: Implementations of the present specification provide a method and an apparatus for extending question and answer samples. According to the method, a random number is generated for each existing sample, a question is blurred for a sample whose random number belongs to sample extension random numbers, to generate an extended sample, so that an overall sample blurring extension rate can be effectively controlled. In addition, for a sample needing blurring extension, a question is extended by deleting a word with a predetermined part of speech in the corresponding question, and then an extended sample is generated based on an extended question, so that more question expression ways are compatible. As such, a question and answer model is trained by using a sample set to which extended samples are added, so that an answer can be provided to a user more effectively.