Affinda's ID Extractor solution can easily be used via API and embedded into existing workflows and platforms.

Process

A typical process that is followed to extract and validate the data from a file is:

End-user submits the file for verification
Customer sends the file to Affinda API for data extraction
Affinda applies the AI model to the file and extracts data
Data is returned via API, along with relevant metadata
Customer uses some rules or logic to determine if human validation is required
If human validation is required:
- reviewUrl found in API response is opened and validation occurs via interface
- Data is retrieved from Affinda again either via API or webhook

Using the API

Affinda's API for the ID Extractor is found at: https://api.affinda.com/v2/custom_documents

Submitting an ID to the Affinda API

The following fields may be included in the API POST request to Affinda. Note, all fields are optional, however, one of file or URL must be included.

Field Name	Description
file	File as binary data blob. Supported formats: PDF, DOC, DOCX, TXT, RTF, HTML, PNG, JPG
url	URL to a file to download and process
file_name	Optional filename of the file
identifier	Unique identifier for the document. If creating a document and left blank, one will be automatically generated.
wait	If "true" (default), will return a response only after processing has completed. If "false", will return an empty data object which can be polled at the GET endpoint until processing is complete.
expiry_time	The date/time in ISO-8601 format when the document will be automatically deleted. Defaults to no expiry.
document_type	The type of ID that has been submitted. Options are passports, aus_drivers_licence, or aus_birth_certificate. If not specified, Affinda will apply classification of what type of ID has been submitted and apply the relevant model to extract data.
reject_duplicates	If "true", parsing will fail when the uploaded file is a duplicate of an existing file. If "false", will parse the document normally whether it is a duplicate or not. If blank, will revert to account level settings.

Response from the Affinda API

The response from the Affinda API will include the data extracted from each field, as well as a range of metadata for each field and the document itself. If wait = false in the POST request, the data will be automatically returned. Otherwise, submit a GET request with the identifier in the body to return the data once extraction is complete.

Field level metadata

Meta Data	Description
rectangle	x/y coordinates for the rectangular bounding box containing the data
confidence	Value that indicates the confidence that the model has that the data returned is correct
isVerified	Indicates whether the data has been validated, either by a human using our validation tool or through auto-validation rules
isClientVerified	Indicates whether the data has been validated by a human
isAutoVerified	Indicates whether the data has been auto-validated
raw	Raw data extracted from the before any post-processing
parsed	Parsed data extract after post-processing steps, including reformatting or mapping to a defined taxonomy
classification	The name of the field

Document level metadata

Metadata	Description
identifier	Unique identifier for the document. If creating a document and left blank, one will be
fileName	Optional filename of the file
ready	If true, the document has finished processing. Particularly useful if an endpoint request specified wait=False, when polling use this variable to determine when to stop polling
readyDt	The datetime when the document was ready
failed	If true, some exception was raised during processing. Check the 'error' field of the main return object.
expiryTime	The date/time in ISO-8601 format when the document will be automatically deleted. Defaults to no expiry.
language	The document's language.
pdf	The URL to the document's pdf (if the uploaded document is not already pdf, it's converted to pdf as part of the parsing process).
parentDocument	Unique identifier for the parent document if the document has been split
childDocuments	Unique identifier for any child documents if the document has been split
pages	The document's pages.
isVerified	This is true if the "confirm" button has been clicked in the Affinda validation tool
reviewUrl	Signed URL (valid for 60 minutes) to access the validation tool. Not applicable for documents types such a resumes.
ocrConfidence	The overall confidence in the conversion of image to text (only applicable for images or PDF documents without a text layer)