All Products
Search

This Product

    Data Management:Knowledge base management

    Document Center

    Data Management:Knowledge base management

    Last Updated:Mar 30, 2026

    Accurate natural language to SQL (NL2SQL) responses depend on Data Copilot having enough context about your business — your terminology, metrics, and common query patterns. The knowledge base is where you store that context. By curating SQL examples and domain-specific fragments, you give Data Copilot what it needs to generate correct SQL and make better automated decisions.

    How the knowledge base works

    Data Management (DMS) builds the knowledge base from three sources:

    Source How it works
    Copilot feedback When you like a Copilot response, DMS automatically saves the SQL statement and related question as a knowledge record (type: SQL Records, level: To Be Supplemented). When you provide feedback or submit a modified SQL statement, DMS saves it as a knowledge record (type: Fragment Knowledge, level: To Be Supplemented).
    Manual addition Add knowledge records directly on the Knowledge Granularity tab.
    Data augmentation pipeline DMS extracts historical SQL statements, converts them into SQL templates, translates those templates into questions, and saves the results as SQL records.

    The Knowledge Granularity tab displays all knowledge records for the current DMS tenant, including both automatically generated and manually added records.

    Knowledge types

    Type Description
    SQL Records SQL statements paired with natural language questions.
    Fragment Knowledge Domain-specific context such as business rules, terminology, and metric definitions.

    Records generated from Copilot feedback start at the To Be Supplemented knowledge level, indicating they are awaiting review.

    Add knowledge records

    Add a single record

    1. Log on to the DMS console V5.0.

    2. Move the pointer over the 2023-01-28_15-57-17.png icon in the upper-left corner and choose All functions > Data Assets > Knowledge Base.

      Note

      In normal mode, choose Data Assets > Knowledge Base in the top navigation bar.

    3. On the Knowledge Granularity tab, click New.

    4. In the Add Knowledge panel, fill in the following fields:

      Field Description
      Description A short label for the knowledge record
      Content The SQL statement or text fragment
      Knowledge Type Select SQL Records or Fragment Knowledge
      Knowledge Level The review status of the record
      Associated Database The database this knowledge applies to

    5. Click OK.

    Import records in bulk

    Only CSV files are supported.

    1. On the Knowledge Granularity tab, click Batch Import.

    2. Download the CSV template, fill in the knowledge information as instructed, and upload the completed file.

    3. Click OK.

    Manage knowledge records

    On the Knowledge Granularity tab, you can edit and delete existing knowledge records.

    Add tags to a knowledge record

    Tags let you filter and locate records quickly.

    Create a tag

    1. On the Knowledge Granularity tab, click Tag Management above the record list.

    2. Click New. In the Create Tag dialog box, enter a tag name and click OK.

      Note

      You can also modify or delete tags.

    Bind a tag to a record

    1. Find the target knowledge record, move the pointer over the image icon, and click Bind.

      image

    2. In the Label dialog box, click Add.

    3. Select one or more tags and click OK.

    Limits

    Item Limit
    Tags per knowledge record 20
    Bulk import format CSV only

    Search for knowledge records

    Filter records by keyword, tag, knowledge level, knowledge type, or associated database.

    image

    What's next

    • DMS Data Copilot — Learn how Copilot uses knowledge records to answer queries and how feedback automatically enriches the knowledge base.