langchain/libs/experimental/langchain_experimental/data_anonymizer/presidio.py

455 lines
16 KiB
Python
Raw Normal View History

Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
from __future__ import annotations
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
import json
from pathlib import Path
from typing import TYPE_CHECKING, Callable, Dict, List, Optional, Union
Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
import yaml
from langchain_experimental.data_anonymizer.base import (
DEFAULT_DEANONYMIZER_MATCHING_STRATEGY,
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
AnonymizerBase,
ReversibleAnonymizerBase,
)
from langchain_experimental.data_anonymizer.deanonymizer_mapping import (
DeanonymizerMapping,
MappingDataType,
create_anonymizer_mapping,
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
)
from langchain_experimental.data_anonymizer.deanonymizer_matching_strategies import (
exact_matching_strategy,
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
)
Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
from langchain_experimental.data_anonymizer.faker_presidio_mapping import (
get_pseudoanonymizer_mapping,
)
if TYPE_CHECKING:
from presidio_analyzer import AnalyzerEngine, EntityRecognizer
from presidio_analyzer.nlp_engine import NlpEngineProvider
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
from presidio_anonymizer import AnonymizerEngine
Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
from presidio_anonymizer.entities import OperatorConfig
def _import_analyzer_engine() -> "AnalyzerEngine":
try:
from presidio_analyzer import AnalyzerEngine
except ImportError as e:
raise ImportError(
"Could not import presidio_analyzer, please install with "
"`pip install presidio-analyzer`. You will also need to download a "
"spaCy model to use the analyzer, e.g. "
"`python -m spacy download en_core_web_lg`."
) from e
return AnalyzerEngine
def _import_nlp_engine_provider() -> "NlpEngineProvider":
try:
from presidio_analyzer.nlp_engine import NlpEngineProvider
except ImportError as e:
raise ImportError(
"Could not import presidio_analyzer, please install with "
"`pip install presidio-analyzer`. You will also need to download a "
"spaCy model to use the analyzer, e.g. "
"`python -m spacy download en_core_web_lg`."
) from e
return NlpEngineProvider
def _import_anonymizer_engine() -> "AnonymizerEngine":
try:
from presidio_anonymizer import AnonymizerEngine
except ImportError as e:
raise ImportError(
"Could not import presidio_anonymizer, please install with "
"`pip install presidio-anonymizer`."
) from e
return AnonymizerEngine
def _import_operator_config() -> "OperatorConfig":
try:
from presidio_anonymizer.entities import OperatorConfig
except ImportError as e:
raise ImportError(
"Could not import presidio_anonymizer, please install with "
"`pip install presidio-anonymizer`."
) from e
return OperatorConfig
Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
# Configuring Anonymizer for multiple languages
# Detailed description and examples can be found here:
# langchain/docs/extras/guides/privacy/multi_language_anonymization.ipynb
DEFAULT_LANGUAGES_CONFIG = {
# You can also use Stanza or transformers library.
# See https://microsoft.github.io/presidio/analyzer/customizing_nlp_models/
"nlp_engine_name": "spacy",
"models": [
{"lang_code": "en", "model_name": "en_core_web_lg"},
# {"lang_code": "de", "model_name": "de_core_news_md"},
# {"lang_code": "es", "model_name": "es_core_news_md"},
# ...
# List of available models: https://spacy.io/usage/models
],
}
Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
class PresidioAnonymizerBase(AnonymizerBase):
Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
def __init__(
self,
analyzed_fields: Optional[List[str]] = None,
operators: Optional[Dict[str, OperatorConfig]] = None,
languages_config: Dict = DEFAULT_LANGUAGES_CONFIG,
add_default_faker_operators: bool = True,
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
faker_seed: Optional[int] = None,
Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
):
"""
Args:
analyzed_fields: List of fields to detect and then anonymize.
Defaults to all entities supported by Microsoft Presidio.
operators: Operators to use for anonymization.
Operators allow for custom anonymization of detected PII.
Learn more:
https://microsoft.github.io/presidio/tutorial/10_simple_anonymization/
languages_config: Configuration for the NLP engine.
First language in the list will be used as the main language
in self.anonymize(...) when no language is specified.
Learn more:
https://microsoft.github.io/presidio/analyzer/customizing_nlp_models/
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
faker_seed: Seed used to initialize faker.
Defaults to None, in which case faker will be seeded randomly
and provide random values.
Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
"""
OperatorConfig = _import_operator_config()
AnalyzerEngine = _import_analyzer_engine()
NlpEngineProvider = _import_nlp_engine_provider()
AnonymizerEngine = _import_anonymizer_engine()
Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
self.analyzed_fields = (
analyzed_fields
if analyzed_fields is not None
else list(get_pseudoanonymizer_mapping().keys())
)
if add_default_faker_operators:
self.operators = {
Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
field: OperatorConfig(
operator_name="custom", params={"lambda": faker_function}
)
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
for field, faker_function in get_pseudoanonymizer_mapping(
faker_seed
).items()
Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
}
else:
self.operators = {}
if operators:
self.add_operators(operators)
provider = NlpEngineProvider(nlp_configuration=languages_config)
nlp_engine = provider.create_engine()
self.supported_languages = list(nlp_engine.nlp.keys())
self._analyzer = AnalyzerEngine(
supported_languages=self.supported_languages, nlp_engine=nlp_engine
)
Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
self._anonymizer = AnonymizerEngine()
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
def add_recognizer(self, recognizer: EntityRecognizer) -> None:
"""Add a recognizer to the analyzer
Args:
recognizer: Recognizer to add to the analyzer.
"""
self._analyzer.registry.add_recognizer(recognizer)
self.analyzed_fields.extend(recognizer.supported_entities)
def add_operators(self, operators: Dict[str, OperatorConfig]) -> None:
"""Add operators to the anonymizer
Args:
operators: Operators to add to the anonymizer.
"""
self.operators.update(operators)
class PresidioAnonymizer(PresidioAnonymizerBase):
def _anonymize(
self,
text: str,
language: Optional[str] = None,
allow_list: Optional[List[str]] = None,
) -> str:
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
"""Anonymize text.
Each PII entity is replaced with a fake value.
Each time fake values will be different, as they are generated randomly.
PresidioAnonymizer has no built-in memory -
so it will not remember the effects of anonymizing previous texts.
>>> anonymizer = PresidioAnonymizer()
>>> anonymizer.anonymize("My name is John Doe. Hi John Doe!")
'My name is Noah Rhodes. Hi Noah Rhodes!'
>>> anonymizer.anonymize("My name is John Doe. Hi John Doe!")
'My name is Brett Russell. Hi Brett Russell!'
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
Args:
text: text to anonymize
language: language to use for analysis of PII
If None, the first (main) language in the list
of languages specified in the configuration will be used.
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
"""
if language is None:
language = self.supported_languages[0]
if language not in self.supported_languages:
raise ValueError(
f"Language '{language}' is not supported. "
f"Supported languages are: {self.supported_languages}. "
"Change your language configuration file to add more languages."
)
2023-10-17 17:27:29 +00:00
# Check supported entities for given language
# e.g. IT_FISCAL_CODE is not supported for English in Presidio by default
# If you want to use it, you need to add a recognizer manually
supported_entities = []
for recognizer in self._analyzer.get_recognizers(language):
recognizer_dict = recognizer.to_dict()
supported_entities.extend(
[recognizer_dict["supported_entity"]]
if "supported_entity" in recognizer_dict
else recognizer_dict["supported_entities"]
)
entities_to_analyze = list(
set(supported_entities).intersection(set(self.analyzed_fields))
)
analyzer_results = self._analyzer.analyze(
Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
text,
2023-10-17 17:27:29 +00:00
entities=entities_to_analyze,
language=language,
allow_list=allow_list,
Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
)
filtered_analyzer_results = (
self._anonymizer._remove_conflicts_and_get_text_manipulation_data(
analyzer_results
)
)
anonymizer_results = self._anonymizer.anonymize(
Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
text,
analyzer_results=analyzer_results,
Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
operators=self.operators,
)
anonymizer_mapping = create_anonymizer_mapping(
text,
filtered_analyzer_results,
anonymizer_results,
)
return exact_matching_strategy(text, anonymizer_mapping)
Add data anonymizer (#9863) ### Description The feature for anonymizing data has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. Anonynization consists of two steps: 1. **Identification:** Identify all data fields that contain personally identifiable information (PII). 2. **Replacement**: Replace all PIIs with pseudo values or codes that do not reveal any personal information about the individual but can be used for reference. We're not using regular encryption, because the language model won't be able to understand the meaning or context of the encrypted data. We use *Microsoft Presidio* together with *Faker* framework for anonymization purposes because of the wide range of functionalities they provide. The full implementation is available in `PresidioAnonymizer`. ### Future works - **deanonymization** - add the ability to reverse anonymization. For example, the workflow could look like this: `anonymize -> LLMChain -> deanonymize`. By doing this, we will retain anonymity in requests to, for example, OpenAI, and then be able restore the original data. - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-08-30 17:39:44 +00:00
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
class PresidioReversibleAnonymizer(PresidioAnonymizerBase, ReversibleAnonymizerBase):
def __init__(
self,
analyzed_fields: Optional[List[str]] = None,
operators: Optional[Dict[str, OperatorConfig]] = None,
languages_config: Dict = DEFAULT_LANGUAGES_CONFIG,
add_default_faker_operators: bool = True,
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
faker_seed: Optional[int] = None,
):
super().__init__(
analyzed_fields,
operators,
languages_config,
add_default_faker_operators,
faker_seed,
)
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
self._deanonymizer_mapping = DeanonymizerMapping()
@property
def deanonymizer_mapping(self) -> MappingDataType:
"""Return the deanonymizer mapping"""
return self._deanonymizer_mapping.data
@property
def anonymizer_mapping(self) -> MappingDataType:
"""Return the anonymizer mapping
This is just the reverse version of the deanonymizer mapping."""
return {
key: {v: k for k, v in inner_dict.items()}
for key, inner_dict in self.deanonymizer_mapping.items()
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
}
def _anonymize(
self,
text: str,
language: Optional[str] = None,
allow_list: Optional[List[str]] = None,
) -> str:
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
"""Anonymize text.
Each PII entity is replaced with a fake value.
Each time fake values will be different, as they are generated randomly.
At the same time, we will create a mapping from each anonymized entity
back to its original text value.
Thanks to the built-in memory, all previously anonymised entities
will be remembered and replaced by the same fake values:
>>> anonymizer = PresidioReversibleAnonymizer()
>>> anonymizer.anonymize("My name is John Doe. Hi John Doe!")
'My name is Noah Rhodes. Hi Noah Rhodes!'
>>> anonymizer.anonymize("My name is John Doe. Hi John Doe!")
'My name is Noah Rhodes. Hi Noah Rhodes!'
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
Args:
text: text to anonymize
language: language to use for analysis of PII
If None, the first (main) language in the list
of languages specified in the configuration will be used.
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
"""
if language is None:
language = self.supported_languages[0]
if language not in self.supported_languages:
raise ValueError(
f"Language '{language}' is not supported. "
f"Supported languages are: {self.supported_languages}. "
"Change your language configuration file to add more languages."
)
2023-10-17 17:27:29 +00:00
# Check supported entities for given language
# e.g. IT_FISCAL_CODE is not supported for English in Presidio by default
# If you want to use it, you need to add a recognizer manually
supported_entities = []
for recognizer in self._analyzer.get_recognizers(language):
recognizer_dict = recognizer.to_dict()
supported_entities.extend(
[recognizer_dict["supported_entity"]]
if "supported_entity" in recognizer_dict
else recognizer_dict["supported_entities"]
)
entities_to_analyze = list(
set(supported_entities).intersection(set(self.analyzed_fields))
)
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
analyzer_results = self._analyzer.analyze(
text,
2023-10-17 17:27:29 +00:00
entities=entities_to_analyze,
language=language,
allow_list=allow_list,
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
)
filtered_analyzer_results = (
self._anonymizer._remove_conflicts_and_get_text_manipulation_data(
analyzer_results
)
)
anonymizer_results = self._anonymizer.anonymize(
text,
analyzer_results=analyzer_results,
operators=self.operators,
)
new_deanonymizer_mapping = create_anonymizer_mapping(
text,
filtered_analyzer_results,
anonymizer_results,
is_reversed=True,
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
)
self._deanonymizer_mapping.update(new_deanonymizer_mapping)
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
return exact_matching_strategy(text, self.anonymizer_mapping)
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
def _deanonymize(
self,
text_to_deanonymize: str,
deanonymizer_matching_strategy: Callable[
[str, MappingDataType], str
] = DEFAULT_DEANONYMIZER_MATCHING_STRATEGY,
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
) -> str:
"""Deanonymize text.
Each anonymized entity is replaced with its original value.
This method exploits the mapping created during the anonymization process.
Args:
text_to_deanonymize: text to deanonymize
deanonymizer_matching_strategy: function to use to match
anonymized entities with their original values and replace them.
"""
if not self._deanonymizer_mapping:
raise ValueError(
"Deanonymizer mapping is empty.",
"Please call anonymize() and anonymize some text first.",
)
text_to_deanonymize = deanonymizer_matching_strategy(
text_to_deanonymize, self.deanonymizer_mapping
)
return text_to_deanonymize
def reset_deanonymizer_mapping(self) -> None:
"""Reset the deanonymizer mapping"""
self._deanonymizer_mapping = DeanonymizerMapping()
Data deanonymization (#10093) ### Description The feature for pseudonymizing data with ability to retrieve original text (deanonymization) has been implemented. In order to protect private data, such as when querying external APIs (OpenAI), it is worth pseudonymizing sensitive data to maintain full privacy. But then, after the model response, it would be good to have the data in the original form. I implemented the `PresidioReversibleAnonymizer`, which consists of two parts: 1. anonymization - it works the same way as `PresidioAnonymizer`, plus the object itself stores a mapping of made-up values to original ones, for example: ``` { "PERSON": { "<anonymized>": "<original>", "John Doe": "Slim Shady" }, "PHONE_NUMBER": { "111-111-1111": "555-555-5555" } ... } ``` 2. deanonymization - using the mapping described above, it matches fake data with original data and then substitutes it. Between anonymization and deanonymization user can perform different operations, for example, passing the output to LLM. ### Future works - **instance anonymization** - at this point, each occurrence of PII is treated as a separate entity and separately anonymized. Therefore, two occurrences of the name John Doe in the text will be changed to two different names. It is therefore worth introducing support for full instance detection, so that repeated occurrences are treated as a single object. - **better matching and substitution of fake values for real ones** - currently the strategy is based on matching full strings and then substituting them. Due to the indeterminism of language models, it may happen that the value in the answer is slightly changed (e.g. *John Doe* -> *John* or *Main St, New York* -> *New York*) and such a substitution is then no longer possible. Therefore, it is worth adjusting the matching for your needs. - **Q&A with anonymization** - when I'm done writing all the functionality, I thought it would be a cool resource in documentation to write a notebook about retrieval from documents using anonymization. An iterative process, adding new recognizers to fit the data, lessons learned and what to look out for ### Twitter handle @deepsense_ai / @MaksOpp --------- Co-authored-by: MaksOpp <maks.operlejn@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
2023-09-07 04:33:24 +00:00
def save_deanonymizer_mapping(self, file_path: Union[Path, str]) -> None:
"""Save the deanonymizer mapping to a JSON or YAML file.
Args:
file_path: Path to file to save the mapping to.
Example:
.. code-block:: python
anonymizer.save_deanonymizer_mapping(file_path="path/mapping.json")
"""
save_path = Path(file_path)
if save_path.suffix not in [".json", ".yaml"]:
raise ValueError(f"{save_path} must have an extension of .json or .yaml")
# Make sure parent directories exist
save_path.parent.mkdir(parents=True, exist_ok=True)
if save_path.suffix == ".json":
with open(save_path, "w") as f:
json.dump(self.deanonymizer_mapping, f, indent=2)
elif save_path.suffix == ".yaml":
with open(save_path, "w") as f:
yaml.dump(self.deanonymizer_mapping, f, default_flow_style=False)
def load_deanonymizer_mapping(self, file_path: Union[Path, str]) -> None:
"""Load the deanonymizer mapping from a JSON or YAML file.
Args:
file_path: Path to file to load the mapping from.
Example:
.. code-block:: python
anonymizer.load_deanonymizer_mapping(file_path="path/mapping.json")
"""
load_path = Path(file_path)
if load_path.suffix not in [".json", ".yaml"]:
raise ValueError(f"{load_path} must have an extension of .json or .yaml")
if load_path.suffix == ".json":
with open(load_path, "r") as f:
loaded_mapping = json.load(f)
elif load_path.suffix == ".yaml":
with open(load_path, "r") as f:
loaded_mapping = yaml.load(f, Loader=yaml.FullLoader)
self._deanonymizer_mapping.update(loaded_mapping)