ZINK (Zero-shot Ink) is a robust Python package designed for the zero-shot anonymization of entities within unstructured text data. In an era where safeguarding sensitive information is critical, ZINK offers an efficient method for redacting or replacing personal details, including names, ages, phone numbers, and medical conditions, while preserving the overall utility of the text for analysis and processing.

Key Features

• Zero-shot Anonymization: No need for training data or pre-trained models, simplifying the anonymization process.
• Flexible Entity Labeling: Users can anonymize various entity types by providing custom labels, catering to unique data needs.
• Redaction and Replacement Options: ZINK allows for two main operations: redacting entities (replacing them with a placeholder format like [LABEL]_REDACTED) or substituting them with generic placeholders.
• Seamless Integration: The package provides a straightforward API, making it easy to incorporate into existing Python projects.

Usage Examples

Redacting Entities

Utilize the redact function to replace identified entities with [LABEL]_REDACTED.

import zink as pss

text = "John works as a doctor and plays football after work and drives a toyota."
labels = ("person", "profession", "sport", "car")
result = pss.redact(text, labels)
print(result.anonymized_text)

Example output:
person_REDACTED works as a profession_REDACTED and plays sport_REDACTED after work and drives a car_REDACTED.

Replacing Entities

Employ the replace function to substitute identified entities with random entities of the same type.

import zink as pss

text = "John Doe dialled his mother at 992-234-3456 and then went out for a walk."
labels = ("person", "phone number", "relationship")
result = pss.replace(text, labels)
print(result.anonymized_text)

Example output:
Warren Buffet dialled his Uncle at 2347789287 and then went out for a walk.

Another Replacement Example

import zink as pss

text = "Patient, 33 years old, was admitted with a chest pain"
labels = ("age", "medical condition")
result = pss.replace(text, labels)
print(result.anonymized_text)

Example output:
Patient, 78 years old, was admitted with a Diabetes Mellitus.

ZINK stands out as a user-friendly and flexible solution for text anonymization, making it a valuable tool for developers and data scientists focused on data privacy.