Detect PII

Parameters

Listed below are the parameters that can be used with this Mantium Transform:

• Source Column: The column name containing the text you want to check for PII. Defaults to content.
• Destination Column: The column name that will hold the PII text. This is a required field. Defaults to PII_detected.
• LLM Model: The large language model used for PII Detection in text. Defaults to gpt-turbo-3.5.
• Prompt Template: The template that guides the identification of PII in text. Defaults to You can only respond with the word ""True"" or ""False"", where your answer indicates whether the text in the user's message contains PII. Do not explain your answer, and do not use punctuation. Your task is to identify whether the text extracted from your company files contains sensitive PII information that should not be shared with the broader company. Here are some things to look out for: - An email address that identifies a specific person in either the local-part or the domain - The postal address of a private residence (must include at least a street name) - The postal address of a public place (must include either a street name or business name) - Notes about hiring decisions with mentioned names of candidates. The user will send a document for you to analyze. Document: $field_source_column.
• Credential ID: The connector from your Mantium account. This is a required field.

Usage

To use the PII Detection transformation, you will need to have a valid API key configured in Mantium for the third-party service (e.g., OpenAI) you want to use. If you don't have one, see the guide here

To use this Mantium Enrichment, follow these steps:

1. Specify the Source Column parameter with the name of the column that contains the text to detect if PII is present.
2. Specify the Destination Column parameter with the name of the column that will hold the True or False value returned to indicate if PII is detected.
3. Specify the LLM Model parameter with the name of the Large Language Model to use to detect for PII.
4. Specify the Prompt Template parameter if desired.
5. Specify the Credential ID parameter with the connector from your Mantium account.
6. Run the transformation by clicking the Save and Run Transforms button. The resulting dataset will have a new column with the specified name containing the PII result.

Example 1: Customer support inquiries

Suppose you have an email inbox for customers to submit feedback and inquiries and you want to flag any containing PII.

Parameters (YAML):

Use the parameters as the configuration in the Mantium app

transform:
  name: Detect PII
  parameters:
    source_column: body
    destination_column: PII_detected
    llm_model: gpt-turbo-3.5
    prompt_template: "You can only respond with the word ""True"" or ""False"", where your answer indicates whether the text in the user's message contains PII. Do not explain your answer, and do not use punctuation. Your task is to identify whether the text extracted from your company files contains sensitive PII information that should not be shared with the broader company. Here are some things to look out for: - An email address that identifies a specific person in either the local-part or the domain - The postal address of a private residence (must include at least a street name) - The postal address of a public place (must include either a street name or business name) - Notes about hiring decisions with mentioned names of candidates. The user will send a document for you to analyze. Document: $field_source_column"
    credential_id: your_credential_id_here

Expected Result Dataset: