All Products
Search
Document Center

Drive and Photo Service:Enable and manage a knowledge base

Last Updated:Feb 27, 2026

The knowledge base in Drive and Photo Service uses the Qwen large language model (LLM) to power enterprise knowledge management. It provides automatic file classification, conversational search, conversational reading, and Q&A capabilities.

Prerequisites

Before you begin, make sure that you have:

Important

This feature is available only in the China (Beijing) and Singapore regions.

Enable the knowledge base

  1. Log on to CDE and go to the admin console.

  2. In the navigation pane, choose Enterprise Settings > AI Toolkit.

  3. On the AI Toolkit page, turn on the Knowledge Base switch.

image

Create a knowledge base

  1. In the navigation pane, choose Knowledge Base, and then click Create First Knowledge Base. image

  2. Configure the following settings: image

    SettingDescription
    Knowledge Base NameEnter a custom name and select an icon. The name cannot exceed 128 characters.
    Knowledge base descriptionEnter a description to help identify the knowledge base. The description cannot exceed 1,024 characters.
    Select associated files for the knowledge baseSelect the scope of files to associate with the knowledge base.
    Specify the associated file typeSelect or customize file types to associate. Currently, only documents and images are supported.
    TeamSelect a team to restrict access to team members. The default permission is Knowledge Base Viewer. You can modify team members and permissions later.
    Default permissions for team membersSet the default permission level: Viewer or Manager.

Manage authorization

  1. On the homepage of the target knowledge base, click Authorization Management.

  2. Set knowledge base permissions for a team or user.

    PermissionDescription
    Knowledge Base ViewerAll users in the specified team, or the specified user, can view the knowledge base.
    Knowledge Base ManagerAll users in the specified team, or the specified user, can manage the knowledge base.
image

Edit a knowledge base

  1. On the homepage of the target knowledge base, click Edit Knowledge Base.

  2. Modify the knowledge base information as needed.

image

Delete a knowledge base

Important

Deleting a knowledge base is permanent and cannot be undone. The source files stored in the drive are not affected.

On the homepage of the target knowledge base, click Delete Knowledge Base.

image

Create a knowledge classification

Knowledge classifications group related files within a knowledge base using AI-powered keyword matching.

  1. On the homepage of the target knowledge base, click Add Sub-knowledge Classification. If no classification exists yet, click Create First Knowledge Classification.

  2. Enter the Subject Keywords and Knowledge Classification Name, and then click Create.

    • Subject Keywords: Enter keywords separated by commas, or enter a natural language phrase. The system uses this input to intelligently group related files into the knowledge base.

    • Knowledge Classification Name: Enter a custom name based on the knowledge base content.

  3. Click a knowledge classification on the left to view the files in it.

imageimage

Manage knowledge classifications

The following operations are available for knowledge classifications:

OperationDescription
Add Sub-knowledge ClassificationCreate sub-classifications to refine the knowledge structure.
Edit knowledge classificationModify the name or properties of a classification.
Authorization ManagementSet access or management permissions for different users.
Delete knowledge classificationRemove classifications that are no longer in use. Use this operation with caution.
AI AssistantUse intelligent search to locate and retrieve information.
Add files to classificationAdd files to a specified knowledge classification.
DownloadExport files or content within a classification for backup or sharing.
image

Keyword guidelines

To improve the accuracy of intelligent classification and retrieval, follow these guidelines when setting Subject Keywords.

Image keywords

The knowledge base uses AI image recognition to extract objective entities from images and use them as keywords for classification, retrieval, and management.

Supported keyword types:

  • Natural landscapes: glaciers, forests, mountains, rivers

  • Living things: pets (cats, dogs), plants, wild animals

  • Man-made objects: buildings, vehicles, electronics, tools

Unsupported keyword types:

  • Abstract or descriptive words: Subjective adjectives or long descriptions, such as "beautiful", "spectacular", or "ancient history"

  • Non-visual concepts: Abstract semantics that cannot be mapped to image content, such as "environmental protection concept", "teamwork spirit", or "future trends"

  • Plain text: Text within images and image filenames

Document keywords

The knowledge base uses semantic matching to calculate relevance between document content and a main topic. This enables intelligent classification, retrieval, and management.

Keyword length

Use phrases of 5 to 20 characters to express a clear, single subject.

Examples: Machine learning model optimization, Supply chain management solution, Product requirements document (PRD)

Keyword combination

Each group can contain multiple related keywords separated by a comma (,).

Example: Employee onboarding and offboarding management, Performance appraisal system, Compensation and bonus management

Keywords to avoid

  • Overly short or generic words Avoid: "AI", "report", "tech-related stuff", "some business processes" Better alternatives: "AI model training", "quarterly summary report", "deep learning image segmentation", "marketing promotion plan"

  • Long natural language descriptions Avoid: "Management systems and operational specifications regarding the company's daily accounting, expense reimbursement processes, budget preparation and execution, fund payment approvals, and financial risk control." Better alternative (extract core ideas): "Daily accounting processes and standards, Expense reimbursement processes and policies, Annual budget preparation and execution management, Fund payment approval process, Financial risk control and internal control system."

See also