LLM_MASK

Description

Used to mask sensitive information in text related to specified labels.

Syntax

LLM_MASK([<resource_name>], <text>, <labels>)

Parameters

Parameter	Description
<resource_name>	The specified resource name
<text>	The text that may contain sensitive information
<labels>	Array of labels to be masked, e.g., ARRAY('name', 'phone', 'email')

Return Value

Returns the text with sensitive information masked. The masked parts are replaced with "[MASKED]".

If any input is NULL, returns NULL.

The result is generated by a large language model, so the output may vary.

Examples

SET default_llm_resource = 'resource_name';
SELECT LLM_MASK('Wccccat is a 20-year-old Doris community contributor.', ['name', 'age']) AS Result;

+-----------------------------------------------------+
| Result                                              |
+-----------------------------------------------------+
| [MASKED] is a [MASKED] Doris community contributor. |
+-----------------------------------------------------+

SELECT LLM_MASK('resource_name', 'My email is rarity@example.com and my phone is 123-456-7890',
                ['email', 'phone_num']) AS RESULT;

+-----------------------------------------------+
| RESULT                                        |
+-----------------------------------------------+
| My email is [MASKED] and my phone is [MASKED] |
+-----------------------------------------------+