Skip to main content
Metadata is “data about data” — it provides background such as structure, format, and meaning. This information lets the AI interpret and process your data more accurately, improving query precision and efficiency. AskTable automatically extracts metadata from your database, including database, table, and field information along with field comments. Even without comments, the system generates relevant descriptions. All metadata is collected and stored in AskTable’s Meta Brain to support later queries and analysis.

​
How to add and edit notes

Open the Data Overview tab on the data source detail page, select a table in the left tree navigation, and you can view and edit its table description and field notes on the right:
  • Table description — the blue-bordered block above the field table. Click its edit icon to modify it.
  • Field notes — in the Note column of the field table, hover over a cell and click the edit icon to open the editor. Enter or change the description and save. The editor supports a character count (up to 255) and one-click AI polish.
Metadata helps with queries in several ways: Example 1: Enumerated values Suppose a field records gender, where sex=1 means male and sex=2 means female. When a user asks for data where “gender is male,” the AI uses the metadata to generate the condition sex=1 automatically, without the user needing to spell it out. Example 2: Specific formats Suppose a field records a fund’s star rating using a format from ★ to ★★★★★. If a user asks “how many funds are rated four stars or above,” the metadata helps the AI understand the star format and generate the correct condition:
  • Wrong: select count(*) from fund where star >= 4;
  • Correct: select count(*) from fund where star in ('★★★★','★★★★★')
With metadata, AskTable understands the real meaning of your data and produces more accurate query results.

​
Note sources

The icon on the left of the Note column marks the source of each field note:
  • Database icon — from the original database comment
  • Robot icon — AI-generated
  • Pen icon — human-annotated
Human-annotated notes best capture business semantics and help the AI understand field meaning.

​
Batch updating descriptions

To update many field descriptions at once, use metadata export / import:
  1. On the data overview page, open the ··· menu and choose Export Metadata to save the current field descriptions, field visibility, and training data as a JSON file
  2. Edit the description fields in the JSON file with a text editor
  3. Back on the data overview page, choose Import Metadata to import the modified file
After import, field descriptions and visibility are overwritten, and training data is appended.