Search

CN-121986332-A - Data retrieval method, system, terminal, storage medium and program product

CN121986332ACN 121986332 ACN121986332 ACN 121986332ACN-121986332-A

Abstract

The disclosure provides a data retrieval method, a system, a terminal, a storage medium and a program product, relates to the technical field of data processing, and is used for logically distinguishing customized data from non-customized data, so as to achieve balance between protecting user privacy and improving localized retrieval precision and efficiency. The method is applied to a terminal and comprises the steps of obtaining a retrieval task, responding to the retrieval task comprising custom information, retrieving custom data matched with the custom information from a multi-mode data source by utilizing index information of the pre-built multi-mode data source, wherein the custom data represents multi-mode data related to a target user, determining a retrieval result according to the custom data, and providing the retrieval result to the user.

Inventors

  • Chu Ta
  • ZHANG NING
  • ZHANG HONGLEI
  • QIAO YONG

Assignees

  • 京东方科技集团股份有限公司
  • 北京京东方技术开发有限公司

Dates

Publication Date
20260505
Application Date
20251219

Claims (20)

  1. 1. A data retrieval method, wherein the method is applied to a terminal, the method comprising: Acquiring a retrieval task; Retrieving, in response to the retrieval task including the custom information, custom data matching the custom information from the multi-modal data source using index information of the pre-built multi-modal data source, the custom data representing multi-modal data associated with the target user; And determining a search result according to the customized data, and providing the search result to a user.
  2. 2. The method of claim 1, wherein obtaining a retrieval task comprises: Waking up the intelligent agent, and triggering and displaying an interactive interface of the intelligent agent; and responding to the multi-mode data input by the user on the interactive interface of the intelligent agent, and acquiring the retrieval task.
  3. 3. The method of claim 1, wherein the customization data includes at least one of data characterizing character relationships, trajectory information, and behavioral habits.
  4. 4. The method of claim 3, wherein, Acquiring use information of a user on a terminal and/or position information of the terminal; and determining the custom data according to the use information and/or the position information.
  5. 5. A method according to claim 3, further comprising: In response to obtaining the latest custom data, dynamically adjusting the retrieval weight of each custom data in the multi-mode data source; Wherein a search weight is used to represent a degree of correlation of the custom data and the custom information, the search weight increasing as the degree of correlation increases.
  6. 6. The method of claim 5, wherein dynamically adjusting the retrieval weight of each custom data in the multimodal data source comprises: Acquiring at least one index data of preference, familiarity and interestingness of a target user on each piece of custom data; and dynamically adjusting the retrieval weight of each piece of custom data according to the index data of each piece of custom data and event elements, wherein the event elements comprise at least one of time, place and event content.
  7. 7. The method of claim 1, wherein the multimodal data source includes customized data representing data related to user privacy and non-customized data representing data not related to user privacy.
  8. 8. The method of claim 7, wherein the custom data is stored locally and the non-custom data is stored at a cloud.
  9. 9. The method of claim 1, wherein the index information of the multi-modal data source is constructed by: vectorizing custom data in a multi-mode data source to obtain custom data vectors; constructing auxiliary index information of the customized data according to the customized data vector; and determining index information of the multi-mode data source according to the auxiliary index information.
  10. 10. The method of claim 1, wherein the multimodal data source comprises a local data source, retrieving custom information matched custom data from the multimodal data source, comprising: and in response to the retrieval task including the custom information, retrieving custom data matched with the custom information from the local data source by utilizing index information of the local multimodal data in the local data source.
  11. 11. The method of claim 1, wherein the multimodal data source comprises a cloud data source, the method further comprising: Responding to the retrieval task comprising non-customized information, and retrieving cloud multi-mode data matched with the non-customized information from a cloud data source by utilizing index information of cloud end multi-mode data in the cloud data source; and determining a search result according to the customized data and the cloud multi-mode data.
  12. 12. The method of claim 1, wherein determining a search result from the custom data comprises: responding to the retrieval task including non-customized information, sending the non-customized information to a cloud large model for network retrieval, and receiving network retrieval data of the cloud large model; and determining a search result according to the customized data and the network search data.
  13. 13. The method of claim 12, wherein determining a search result from the customization data and the network search data comprises: retrieving custom data including the network retrieval data feature from the custom data, determining the retrieved custom data as a retrieval result, or, And retrieving network retrieval data containing the custom data features from the network retrieval data, and determining the retrieved network retrieval data as a retrieval result.
  14. 14. The method of claim 1, wherein the multimodal data source comprises a local data source, the method further comprising: In response to the retrieval task including non-customized information, retrieving local multimodal data matching the non-customized information from the local data source using index information of the local multimodal data in the local data source; and determining a retrieval result according to the customized data and the local multi-modal data.
  15. 15. The method of claim 1, wherein determining a search result from the custom data comprises: Responding to the retrieval task comprising non-customized information, sending the non-customized information to a local large model for retrieval, and receiving local retrieval data of the local large model; and determining a search result according to the customized data and the local search data.
  16. 16. The method of claim 1, wherein determining a search result from the custom data comprises: the custom data is determined as a result of the search or, And inputting the customized data and the retrieval task into a local large model, and determining an output result of the local large model as a retrieval result.
  17. 17. The method of claim 1, wherein retrieving custom information matched custom data from a multi-modal data source using the index information comprises: retrieving, from a multi-modal data source, custom data identical to the category using the category of the index information and the custom information; and retrieving the custom data matched with the custom information from the custom data with the same category.
  18. 18. The method of claim 1, wherein retrieving custom information matched custom data from a multi-modal data source using the index information comprises: Extracting key semantic information of the custom information; And retrieving the customized data matched with the key semantic information from a multi-mode data source by utilizing the index information and the key semantic information.
  19. 19. The method of claim 1, wherein the customization information includes image data associated with a specified persona, the customization data including image data of an image library, retrieving customization data matching the customization information from a multimodal data source, comprising: In response to the retrieval task including image data related to a specified person, retrieving the image data related to the specified person from an image library using index information of image data of each category in the image library; Wherein the categories of image data are used to distinguish between different persona relationships.
  20. 20. The method of claim 1, wherein the customization information includes data related to occurrence of a specified behavior, wherein the customization data includes behavior habit data of a user, and wherein retrieving customization data matching the customization information from a multimodal data source includes: and responding to the retrieval task to comprise data related to the appointed behavior, and retrieving the data related to the appointed behavior from the behavior habit data according to index information of the behavior habit data of the user.

Description

Data retrieval method, system, terminal, storage medium and program product Technical Field The present disclosure relates to the field of data processing technologies, and in particular, to a data retrieval method, a system, a terminal, a storage medium, and a program product. Background In the current explosive growth age background of big data, traditional data processing methods have difficulty in meeting increasingly complex demands. For example, the file formats stored in the mobile phone terminals of users are diversified, involving pictures, texts, audio, video, etc., and the volume of part of the files is huge, for example, the pictures and video are often calculated in hundreds of M, which provides a considerable challenge for users to accurately find their personal data. Furthermore, while the current large model application is exhibiting explosive growth, the mobile terminal is limited by insufficient computing power, resulting in the large model being applied in the form of "terminal question-cloud processing-token (minimum semantic or feature unit) feedback". Part of mobile phone manufacturers are gradually developing terminal large models in order to solve the contradiction between protecting personal data privacy of users and utilizing the function of 'doodling strong knowledge' of the large models, but the current retrieval effect is still unsatisfactory. Disclosure of Invention The present disclosure provides a data retrieval method, system, terminal, storage medium, and program product for logically distinguishing between custom data and non-custom data, which balances protection of user privacy and improvement of localized retrieval accuracy and efficiency. In a first aspect, a data retrieval method provided by an embodiment of the present disclosure is applied to a terminal, where the method includes: Acquiring a retrieval task; Retrieving, in response to the retrieval task including the custom information, custom data matching the custom information from the multi-modal data source using index information of the pre-built multi-modal data source, the custom data representing multi-modal data associated with the target user; And determining a search result according to the customized data, and providing the search result to a user. In a second aspect, a data retrieval system provided by an embodiment of the present disclosure includes an interaction layer, an access layer, and a processing layer, where: The interaction layer is used for acquiring a retrieval task, the access layer is used for accessing a multi-mode data source, and the processing layer is configured to execute the following steps: acquiring a multi-mode data source from the access layer, and constructing index information of the multi-mode data source; Retrieving, in response to the retrieval task including the customization information, from a multimodal data source, customization data matching the customization information using the index information, the customization data representing multimodal data associated with the target user; and determining a search result according to the customized data, and providing the search result to a user through the interaction layer. In a third aspect, a data retrieval method provided by an embodiment of the present disclosure is applied to a cloud server, where the method includes: Receiving a search instruction sent by a terminal with access rights; Retrieving, in response to the retrieval instruction including the custom information, custom data matching the custom information from the multi-modal data source using index information of the pre-built multi-modal data source, the custom data representing multi-modal data associated with the target user; and determining a search result according to the customized data, and sending the search result to a terminal. In a fourth aspect, the embodiments of the present disclosure further provide a terminal, the terminal including a processor and a memory, the memory being configured to store a program executable by the processor, the processor being configured to read the program in the memory and perform the steps of: Acquiring a retrieval task; Retrieving, in response to the retrieval task including the custom information, custom data matching the custom information from the multi-modal data source using index information of the pre-built multi-modal data source, the custom data representing multi-modal data associated with the target user; And determining a search result according to the customized data, and providing the search result to a user. In a fifth aspect, embodiments of the present disclosure further provide a cloud server, including a processor and a memory, the memory being configured to store a program executable by the processor, the processor being configured to read the program in the memory and perform the steps of: Receiving a search instruction sent by a terminal with access rights; Retrieving, in response to the r