Dynamic facet generation using large language models

Patent metadata

Field	Value
Publication number	US-12373391-B1
Application number	US-202418787759-A
Country	US
Kind code	B1
Filing date	Jul 29, 2024
Priority date	Jul 29, 2024
Publication date	Jul 29, 2025
Grant date	Jul 29, 2025

Abstract

The present disclosure is directed toward systems, methods, and non-transitory computer-readable media for generating a dynamic facet by using a large language model. For example, the disclosed systems extract raw facet data from a plurality of content items stored in a content management system. In addition, the disclosed systems determine one or more facet content groups by grouping the plurality of content items according to the raw facet data. Further, the disclosed systems generate a facet prompt from the one or more facet content groups. Moreover, the disclosed systems generate a dynamic facet by providing the facet prompt to a large language model.

First claim

What is claimed is:

1. A system comprising:
at least one processor; and
a non-transitory computer-readable medium storing instructions which, when executed by the at least one processor, cause the system to:
prior to a client device searching a content management system and in response to the client device navigating the content management system;
extract raw facet data from a plurality of content items stored in the content management system, wherein navigating the content management system comprises a user of the client device selecting a folder or sub-folder within the content management system;
determine a plurality of facet content groups by grouping the plurality of content items according to the raw facet data;
identify a subset of the plurality of facet content groups based on ranking the plurality of facet content groups according to a threshold range of content items associated with each of the plurality of facet content groups;
generate a facet prompt from the subset of the plurality of facet content groups, wherein the facet prompt includes instructions to a large language model that are created in response to the client device navigating the content management system and prior to the client device performing the search of the content management system;
generate a dynamic facet by providing the facet prompt to the large language model, wherein the facet prompt comprises the subset of the plurality of facet content groups; and

provide the dynamic facet to a graphical user interface of the client device, wherein the dynamic facet is selectable by the client device.

2. The system of claim 1 , further storing instructions which, when executed by the at least one processor, cause the system to extract the raw facet data from the plurality of content items prior to the client device searching the content management system by:
extracting one or more metadata tags for a content item of the plurality of content items comprising a topic of the content item;
extracting access data packets for the content item that indicates one or more client devices that accessed the content item; or
extracting a subfolder location for a content item of the plurality of content items.

3. The system of claim 1 , further storing instructions which, when executed by the at least one processor, cause the system to extract the raw facet data from the plurality of content items prior to the client device searching the content management system by:
extracting one or more word combinations from a file name for a content item of the plurality of content items;
extracting a file type for the content item of the plurality of content items; or
extracting operation data packets for the content item that indicates one or more operations performed on the content item.

4. The system of claim 1 , further storing instructions, which when executed by the at least one processor cause the system to:
generate the dynamic facet by generating the facet prompt that includes instructions to the large language model to abstract groupings from the subset of the plurality of facet content groups,
wherein the dynamic facet is generated as part of pre-processing steps performed prior to the client device searching the content management system.

5. The system of claim 1 , further storing instructions, which when executed by the at least one processor cause the system to:
determine permissions of a client device to access one or more content items associated with the dynamic facet;
determine that the client device does not have permission to access a content item associated with the dynamic facet; and
remove the content item from the dynamic facet.

6. The system of claim 1 , further storing instructions, which when executed by the at least one processor cause the system to:
in response to receiving a selection of the dynamic facet by a client device,
generate one or more additional facet content groups by grouping one or more content items of the dynamic facet according to the raw facet data;
generate an additional facet prompt from the one or more additional facet content groups; and
generate an additional dynamic facet by providing the additional facet prompt to the large language model.

7. A computer-implemented method comprising:
prior to a client device searching a content management system and in response to a client device navigating a content management system, extracting raw facet data from a plurality of content items stored in the content management system;
determining one or more facet content groups by grouping the plurality of content items according to the raw facet data;
generating a facet prompt comprising the one or more facet content groups, wherein the facet prompt includes instructions to a large language model that are created in response to the client device navigating the content management system and prior to the client device performing the search of the content management system; and
generating a dynamic facet by providing the facet prompt to the large language model, wherein the facet prompt comprises a subset of the one or more facet content groups.

8. The computer-implemented method of claim 7 , wherein extracting the raw facet data from the plurality of content items comprises at least one of:
extracting one or more metadata tags for a content item of the plurality of content items comprising a topic of the content item;
extracting access data packets for the content item that indicates one or more client devices that accessed the content item; or
extracting operation data packets for the content item that indicates one or more operations performed on the content item.

9. The computer-implemented method of claim 7 , wherein extracting the raw facet data from the plurality of content items comprises at least one of:
generating, utilizing a machine learning classification model, classes of the plurality of content items; and
extracting the classes of the plurality of content items as the raw facet data.

10. The computer-implemented method of claim 7 , wherein determining the one or more facet content groups further comprises generating a mapping for the one or more facet content groups to the plurality of content items by:
assigning a first content item with a first facet content group based on identifying the first content item of the plurality of content items corresponds to the first facet content group; and
assigning a second content item with a second facet content group based on identifying the second content item of the plurality of content items corresponds to the second facet content group.

11. The computer-implemented method of claim 7 , wherein determining the one or more facet content groups further comprises generating a mapping for the one or more facet content groups to the plurality of content items by:
assigning a first content item with a first facet content group based on identifying the first content item of the plurality of content items that corresponds to the first facet content group;
assigning the first content item with a second facet content group based on identifying the first content item of the plurality of content items corresponds to the second facet content group;
determining a string length of the first facet content group is greater than a string length of the second facet content group; and
based on the string length of the first facet content group, assigning the first content item to the first facet content group and not to the second facet content group.

12. The computer-implemented method of claim 7 , further comprising ranking the one or more facet content groups to identify a subset of facet conten

Assignees

• Dropbox Inc

Inventors

• Liao Wesley
• Wei Ermo
• Dash Aleksander
• Zerby Morgan
• Shubbar Simon
• Upadhyaya Prasang
• Asher Maor Bar
• Ding Jiarui

Classifications

• G06F21/6218

  to a system of files or objects, e.g. local or distributed file system or database · CPC title

• G06F16/168 Primary

  Details of user interfaces specifically adapted to file systems, e.g. browsing and visualisation, 2d or 3d GUIs (query results presentation G06F16/156) · CPC title

• G06F16/164 Primary

  File meta data generation · CPC title