Method and apparatus for data search, system, device and computer readable storage medium

Patent metadata

Field	Value
Publication number	US-11636155-B2
Application number	US-202117169661-A
Country	US
Kind code	B2
Filing date	Feb 8, 2021
Priority date	Sep 8, 2020
Publication date	Apr 25, 2023
Grant date	Apr 25, 2023

Abstract

The present disclosure provides a data search method, and relates to the field of computer technology. The method includes determining semantic-understanding information corresponding to a received search term by subjecting the search term to semantic understanding; analyzing the semantic-understanding information to obtain timeliness requirement information of the search term; determining an acquisition mode of resource result corresponding to the search term based on the timeliness requirement information of the search term; and obtaining the resource result corresponding to the search term by using the determined acquisition mode. The present disclosure further provides a data search apparatus, a system, a device, and a computer readable storage medium.

First claim

What is claimed is:

1. A data search method, comprising:
determining semantic-understanding information corresponding to a received search term by subjecting the search term to semantic understanding;
analyzing the semantic-understanding information to obtain timeliness requirement information of the search term;
determining an acquisition mode of resource result corresponding to the search term based on the timeliness requirement information of the search term; and
obtaining the resource result corresponding to the search term by using the determined acquisition mode,
wherein the acquisition mode of resource result comprises a mode of retrieving an existing resource result from a preset cache and a mode of triggering penetration of the preset cache to obtain an updated resource result from a back end,
wherein the step of analyzing the semantic-understanding information to obtain the timeliness requirement information of the search term comprises:
calculating the timeliness requirement information of the search term according to an information item in the semantic-understanding information and a preset weight coefficient corresponding to the information item in the semantic-understanding information, wherein the semantic-understanding information comprises at least one of the following information items: literal meaning, semantic integrity and probability of burst keyword, and
wherein the step of determining the acquisition mode of resource result corresponding to the search term based on the timeliness requirement information of the search term comprises:
determining, when a calculated value of the timeliness requirement information of the search term is greater than or equal to a preset first score, that the acquisition mode of resource result corresponding to the search term is the mode of triggering penetration of the preset cache to obtain the updated resource result from the back end, and
determining, when the calculated value of the timeliness requirement information of the search term is less than the preset first score, that the acquisition mode of resource result corresponding to the search term is the mode of retrieving the existing resource result from the preset cache.

2. The method of claim 1 , wherein the step of determining the acquisition mode of resource result corresponding to the search term based on the timeliness requirement information of the search term comprises:
determining system information corresponding to the search term;
comprehensively analyzing timeliness and machine cost of the search term according to the timeliness requirement information and the system information of the search term to obtain a comprehensive analysis result of the search term; and
determining the acquisition mode of resource result corresponding to the search term according to the comprehensive analysis result of the search term.

3. The method of claim 2 , wherein the step of comprehensively analyzing the timeliness and the machine cost of the search term according to the timeliness requirement information and the system information of the search term to obtain the comprehensive analysis result of the search term comprises:
calculating the timeliness requirement information of the search term according to an information item in the semantic-understanding information and a preset weight coefficient corresponding to the information item in the semantic-understanding information;
calculating machine cost information of the search term according to an information item in the system information and a preset weight coefficient corresponding to the information item in the system information; and
obtaining the comprehensive analysis result of the search term according to a combination of the timeliness requirement information and the machine cost information of the search term.

4. The method of claim 1 , after calculating the timeliness requirement information of the search term, further comprising:
determining retrieval-related information of the search term comprising at least one of a search frequency of the search term and a total number of recalled results corresponding to the search term;
calculating timeliness-related information of the search term according to the retrieval-related information and a preset weight coefficient corresponding to the retrieval-related information; and
taking the obtained timeliness requirement information and the timeliness-related information of the search term as the timeliness requirement information of the search term.

5. The method of claim 2 , before comprehensively analyzing the timeliness and the machine cost of the search term according to the timeliness requirement information and the system information of the search term, further comprising:
preprocessing the timeliness information and the system information, wherein a preprocess comprises at least one of data normalization and feature standardization.

6. The method of claim 3 , before comprehensively analyzing the timeliness and the machine cost of the search term according to the timeliness requirement information and the system information of the search term, further comprising:
preprocessing the timeliness information and the system information, wherein a preprocess comprises at least one of data normalization and feature standardization.

7. The method of claim 1 , wherein
the preset cache is a cache corresponding to a predetermined back-end process comprising at least one of the following processes: recall-layer process, sorting-layer process, fusion-layer process and access-layer process; and
the preset cache comprises at least one of the following caches: a fusion-layer cache and a sorting-layer cache.

8. The method of claim 7 , wherein
the information item in the system information comprises at least one of system capacity, system response time, system response speed, system load, and total number of resource requests; and
the different back-end processes correspond to different information items in the semantic-understanding information, and also correspond to different information items in the system information, and the weight coefficient corresponding to an information item in the system information varies when the search term is searched for at different times.

9. The method of claim 2 , wherein
the preset cache is a cache corresponding to a predetermined back-end process comprising at least one of the following processes: recall-layer process, sorting-layer process, fusion-layer process and access-layer process; and
the preset cache comprises at least one of the following caches: a fusion-layer cache and a sorting-layer cache.

10. The method of claim 9 , wherein
the information item in the system information comprises at least one of system capacity, system response time, system response speed, system load, and total number of resource requests; and
the different back-end processes correspond to different information items in the semantic-understanding information, and also correspond to different information items in the system information, and the weight coefficient corresponding to an information item in the system information varies when the search term is searched for at different times.

11. The method of claim 3 , wherein
the preset cache is a cache corresponding to a predetermined back-end process comprising at least one of the following processes: recall-layer process, sorting-layer process, fusion-layer process and access-layer process; and
the preset cache comprises at least one of the following caches: a fusion-layer cache and a sorting-layer cache.

12. The method of claim 11 , wherein
the information item in the syste

Assignees

• Baidu online network technology beijing co ltd

Inventors

• Xie Da
• Fan Biao
• Zheng Zhixun
• Chen Jiajie
• Lv Xiaoyong
• Zhen Zhen
• Li Duo
• Sun Chong

Classifications

• G06F16/9038

  Presentation of query results · CPC title

• G06F16/3334

  Selection or weighting of terms from queries, including natural language queries · CPC title

• G06F40/30 Primary

  Semantic analysis · CPC title

• G06F16/24565

  Triggers; Constraints · CPC title

• G06F16/3344 Primary

  using natural language analysis · CPC title