Creating a new Extraction Model
You can create and configure a new Information Extraction model in just minutes—with accurate performance
Step 1: Upload your Document(s)
When you first create a Workspace, you will be prompted to upload your first document(s).
Step 2: Select 'Add New Document Type'
After uploading, you will be prompted to select 'Add New Document'.
If a Document Type has already been added to your Workspace, the classification model may automatically classify the document. To override this, click the dropdown next to the Document Type (above the list of fields in the right-side panel) and select 'Remove Classification'. You can then proceed to 'Add New Document Type'.
Step 3: Automatic Classification
Our AI model will interpret the type of document uploaded and determine what type of document it is.
Step 4: Suggested Fields and Extracted Data
Once the document has been classified, Affinda will then suggest some fields that may be relevant for extraction for that Document Type. The associated data (where present) will be extracted from the document for review.
Step 5: Document Type Configuration
At this stage, users are in the Document Configuration view, where they can review and adjust document classification and field settings.
Document Reclassification
If a document has been misclassified, users can update its classification by renaming the document. Note that for our default classification model, Affinda considers the document name, so it should accurately reflect the type of documents being uploaded.
Field Configuration
The model suggests fields that are likely useful for extraction, but adjustments may be necessary.
- Disabling or Deleting Fields – Users can disable or delete any suggested fields that are not needed.
- To disable a field, toggle the blue switch in the field list or within the Edit Field view. While disabled fields can be re-enabled, the model may not consistently predict them if documents are added to Model Memory without these fields annotated.
- To delete a field, select Delete from the Edit Field view.
- Adding New Fields – Users can add fields from the bottom of the right panel or by selecting a field heading. A prompt will appear to enter a label, choose a field type, and configure additional settings.
- Editing Fields – Users can modify a field’s label, data type, or other settings as needed.
See Field Formatting for more information.
Once the fields are configured, click 'Done'.
Step 6: Confirm extracted data
Users will be sent to the document validation UI where the extracted data should be reviewed for accuracy. When the data is correct, users can click 'Confirm Document' and the document will be added to Model Memory to improve performance
See User Validation for more information on how to validate the extracted data and Model Memory for more information on how Affinda uses validated documents to improve performance over time.
Step 7: Reparse unclassified documents
If users upload multiple documents at the start of the process, Affinda will automatically attempt to reclassify any unclassified documents once the first document is validated. This reduces the need for manual reclassification into the newly created Document Type.
- If a document matches the newly created Document Type, Affinda will extract the relevant data
- If a document does not match the newly created Document Type, users will be prompted to either:
- Reject the document,
- Manually classify, or
- Create a new Document Type (repeat from Step 2 above)
Updated 1 day ago