Finalize an item for training
Background
This document describes how to report back the correct values for an item that was interpreted by Aiida. When this endpoint is called, the item will be moved to the archive column and will be in a read only state allowing the item to be trained by the algorithms. That means that this is a preparation step for the item to be able to be trained. The actual training will happen automatically at a later stage.
Note: This endpoint should only be called once per item.
Note: You need to send back all values that should be trained even those values that have not been changed.
Note: At all costs avoid reporting back of unsupervised(unseen) and non correct values. That means that all values that are sent back should be seen and validated by a human. The values that are not supervised should be ignored and not be sent back in the item finalization step.
Finalize an item
Step 1: Upload an item
Upload an item, and when it is interpreted, retrieve the item including values, classification and accountings (optional) as described elsewhere in the documentation. Check whether any values need to be corrected.
Step 2: Finalize the item
Finalize the item by a call to
Example
{
"values": [
{"field_name": "Supplier Name", "value:": "Acme Inc."},
{"field_id": "153a4d4a-bec2-11eb-bc90-528c3b9e15df", "value": "115.50"}
],
"accounting": [
{
"account": "5410",
"debit": "230.00",
"credit": "0.00",
"dimensions": {"cost_center": "99", "project": "Project 1"},
"transaction_information": "string",
}
],
"classification": "SUPPLIER_INVOICE"
}
Body parameters
Parameter | Comment |
values
|
List of field values (json, optional) |
accounting
|
List of accountings (json, optional) |
classification
|
Classification of item (string, optional). Must match one of the classifications of your workflows. Skip if not applicable. |
Schema for values
field_name
|
Field name (must match an existing field) (optional) |
field_id
|
Field id (must match an existing field) (optional) |
value
|
The correct value (optional) |
One of either field_name or field_id must be present.
A missing value means the same as an empty string, erasing the value.
Schema for accounting
account
|
Account number (4 digits) |
debit
|
Amount for debit (string) |
credit
|
Amount for credit (string) |
dimensions
|
project and/or cost_center (json, optional) |
transaction_information
|
Transaction information (optional) |
Usage
In the request body, include all those values (for values, accountings, and classification) that you have seen and validated (both correct and incorrect).
Example 1: If Aiida reports a value for “Supplier Name”, but nobody on your end has seen and validated the value, then do not include it in the payload.
Example 2: If Aiida reports a value for “VAT”, and you have verified that it was correct, then include it in the payload.
Example 3: If Aiida reports a value for “Total”, and you find that it is incorrect, then include the correct value in the payload.
Example 4: If Aiida did not report a value for “BG”, but there is a correct value, then include the correct value in the payload.
Example 5: If Aiida reported a value for “PG”, but there is no correct value (i.e. the PG field should be empty), then return an empty value for PG in the payload: {“field_name”: “PG”}