Our pattern redaction functionalities are built with an integration to Microsoft Azure AI Services. This service uses both natural language processing and machine learning models for these detections.

⚠️ Good to know

It's essential to highlight that, as it's all unsupervised AI, it’s crucial that any AI pattern suggestions provided by the AI pattern detection and recognition service, alongside any items missed by AI are reviewed by the users or subject experts 🤓.

Patterns available to redact

We now support Personally Identifying Information (PII) patterns by Microsoft and can recognize PII patterns on PDF documents added after Monday, 13th February 2023. If you wish to apply AI patterns redaction to existing documents uploaded before this date, you'll need to re-upload them so they can undergo the recognition process.

Please find the list of supported patterns below:

Pattern Heading	Description
Name	Names of people
Job	Job types or roles held by a person
Organization	Companies, political groups, musical bands, sport clubs, government bodies, and public organizations. Nationalities and religions are not included in this entity type. Sub-category: Medical: Medical companies and groups Stock exchange: Stock exchange groups. Sports: Sports-related organizations.
Street address	Full mailing address
Phone number	Phone numbers (US and EU phone numbers only)
Email	Email addresses
Website	URLs to websites
IP address	Network IP addresses
Date	Dates and times of day Sub-category: Date: Calendar dates Time: Times of day DateRange: Date ranges TimeRange: Time ranges Duration: Durations Set: Set, repeated times
Age	Ages
SWIFT code	SWIFT codes for payment instruction information
Credit card	Credit card numbers.
IBAN	IBAN codes for payment instruction information
Identity	Depending on countries & languages, for example for Australia: Australia bank account number Australian business number Australia Company Number Australia driver's license Australia medical account number Australia passport number Australia tax file number

ℹ️ Note: For more details on the supported patterns and entity categories, please refer to this page.

FAQ 🤔

What languages does the AI pattern support?

Answer: The AI pattern detection and recognition service typically identifies the language using the initial part of the document. Currently, it supports the following languages:

English
French
German
Italian
Japanese
Korean
Portuguese (Brazil)
Portuguese (Portugal)
Spanish

How accurate is the AI pattern recognition?

Answer: The recognized patterns come with a confidence score ranging from 0 to 1. Ansarada filters these outputs to retain only matches with a high level of accuracy.