公开(公告)号：US20250005067A1

公开(公告)日：2025-01-02

申请号：US18745859

申请日：2024-06-17

Applicant: BEIJING BAIDU NETCOM SCIENCE TECHNOLOGY CO., LTD.

Inventor： Yushen Chen ,  Guangyao Han ,  Lei Su ,  Hongda Yue ,  Yi Wang ,  Yunjing An ,  Lin Chang

IPC: G06F16/383 ,  G06F16/33 ,  G06F40/30

Abstract: The disclosure provides a sentence generation method based on a large language model including: obtaining a query sentence, in which the query sentence has at least one candidate description object; performing semantic recognition on the query sentence to obtain first semantic information and second semantic information corresponding to the query sentence, in which categories of the first semantic information and the second semantic information are different; inputting the at least one candidate description object and the first semantic information into the large model, to identify a target description object from the at least one candidate description object based on the large model; selecting target service data from at least one piece of service data corresponding to the target description object based on the second semantic information; and generating a reply sentence corresponding to the query sentence according to the target service data.

公开(公告)号：US12174917B2

公开(公告)日：2024-12-24

申请号：US18169806

申请日：2023-02-15

Applicant: BEIJING BAIDU NETCOM SCIENCE TECHNOLOGY CO., LTD.

Inventor： Shuo Li ,  Hanchenxi Xu ,  Juyan Zhang ,  Hongda Yue ,  Haiyang Xu

IPC: G06F16/00 ,  G06F16/23 ,  G06F18/22

Abstract: A data processing method is provided. The method includes: obtaining a sample data set for modeling; selecting a first sample data from the sample data set; generating, in response to determining that a similarity between a first semantic vector corresponding to a first feature dimension and a second semantic vector corresponding, to a second feature dimension meets a preset condition, a second sample data based on the first sample data; and adding the second sample data to the sample data set.

公开(公告)号：US20230081015A1

公开(公告)日：2023-03-16

申请号：US18050580

申请日：2022-10-28

Applicant: BEIJING BAIDU NETCOM SCIENCE TECHNOLOGY CO., LTD.

Inventor： Yushen Chen ,  Guangyao Han ,  Lei Su ,  Zeqing Jiang ,  Zhiping Li ,  Hongda Yue ,  Haijing Xu ,  Ming Luan

IPC: G06F40/30 ,  G06F40/205 ,  G06V30/148

Abstract: Disclosed are a method for acquiring information. The method includes: acquiring a file to be processed and an information type; recognizing at least one piece of candidate information related to the information type from the file to be processed; determining a target recognition feature and a semantic feature of each piece of candidate information, the target recognition feature is configured to describe a matching condition between the each piece of candidate information and the information type; and determining target information from the at least one piece of candidate information based on the target recognition feature and the semantic feature.