1 AI Assistant with AWS Comprehend

The ATIPXpress AI Assistant now integrates with AWS Comprehend to create powerful AI redaction templates. Designed to simplify the redaction of sensitive information, these templates harness the power of artificial intelligence to automate and enhance the redaction process.

1.1 About AWS Comprehend

AWS Comprehend is a natural language processing service. Using AWS Comprehend-powered templates, you can automatically identify:

Entities: Real-world objects such as people, places, dates (applicable to documents in any language)
Personally Identifiable Information (PII): Textual references to personal data that could be used to identify an individual (Available for English and Spanish language documents)

1.2 About the ATIPXpress AI Assistant

The ATIPXpress AI Assistant uses AI Redaction Templates to consistently identify PII and other sensitive text in documents and highlight or redact automatically. Permissioned users can create and edit templates tuned to locate specific words or phrases, named entities, personal pronouns, and user-defined patterns.

Users in turn can apply these templates to documents, automatically identifying the selected information and highlighting and redacting as defined in the template.

The AI Assistant is an add-on feature for ATIPXpress and requires a license to use.

2 Creating Templates Using AWS Comprehend

To create a redaction template using AWS Comprehend:

Navigate to the AI Redaction Templates configuration (Administration > Document Management > AI Redaction Templates) and click New to create a new template:

From the New AI Redaction Template screen, select the AWS Comprehend radio button:

Note: Templates can only use either Core NFP or AWS Comprehend.

The screen refreshes to show a new set of redaction options for AWS Comprehend. There are sections for (A) Entities (Aws Comprehend) and (B) PII (Aws Comprehend PII):

Use the checkboxes to select items to identify using AWS Comprehend, as well as action (Redaction, Highlight, Strikethrough), redaction code, and color.
Once you’re done configuring the template, click Save at the bottom of the screen to save the new (or edited) template.

See the AI Redaction Templates section of the User & Administrator Manual for a deeper dive into creating templates.

3 What AWS Comprehend Can Do

See the following subsections for details on the type of information AWS Comprehend can locate to assist with document review and redaction.

3.1 Entities

Entities represent real-world objects such as people, places, dates, and more. Each entity is associated with a confidence score indicating the accuracy of its detection. You can use AWS Comprehend to identify and redact entities in any language text document, including the items in the following table:

Entity	Description
Date	Refers to a full date, day, month, or time
Event	Denotes an event like a festival, concert, or election
Location	Identifies a specific place such as a country, city, or building
Organization	Represents large entities like governments, companies, or sports teams
Person	Represents individuals, groups, nicknames, or fictional characters
Quantity	Denotes a quantified amount like currency, percentages, or numbers
Title	Refers to official names given to creations or creative works such as movies, books, or songs
Commercial Item	Represents a branded product
Other	Categorizes entities that don't fit into the other predefined types

For more about Entities, see the following resource from AWS:

https://docs.aws.amazon.com/comprehend/latest/dg/how-entities.html

3.2 PII

You can use the Amazon Comprehend to detect personally identifiable information (PII) in English or Spanish text documents. PII is a textual reference to personal data that could be used to identify an individual. PII examples include addresses, bank account numbers, and phone numbers.

The available PII to add to AI Redaction Templates include:

Entity	Description
Address	Physical address including street, city, state, and country
Email	Email address
Int Bank Account Number	Bank account number format
IP Address	IPv4 address
License Plate	Vehicle license plate number
MAC Address	Network interface controller identifier
Password	Alphanumeric string used as a password
Phone	Phone number including fax and pager numbers
Pin	Four-digit personal identification number
Swift Code	Bank Identifier Code for money transfers
URL	Web address

Entity	Description
Username	User account identifier
Vehicle Identification	Unique vehicle identifier (VIN)
Credit Debit CVV	Three or four-digit card verification code
Credit Debit Expiry	Expiration date of credit or debit cards
Credit Debit Number	Credit or debit card number
Date Time	Date and time information
Driver ID	Driver's license number
Age	Individual's age with quantity and time unit
Name	Individual's name (excluding titles)
Bank Account Num	US bank account number
Bank Routing	US bank account routing number
Passport Number	US passport number
TAX Identification Num	US Individual Taxpayer Identification Number (ITIN)
SSN	US Social Security Number

For more information about PII with AWS Comprehend, see the following materials from AWS:

https://docs.aws.amazon.com/comprehend/latest/dg/pii.html

ATIPXpress AI Assistant with AWS Comprehend