...
The review pane consists of three main building blocks:
The document Document viewThe data
The context bar
The toolbar
...
Finalise the document for download or further processing
Create an additional datapoint for training so that we will make fewer mistakes in predictions going forward.
...
Reviewing a document
Read the steps and/or watch the short walkthrough video underneath.
Go to the inbox pane
Select the file you want to review, click on the chain icon
The file will appear in the review mode of our product
Check which fields were not confidently recognized by our machine learning models and draw boxes around the values it should have recognizedor the original filename. You will be redirected to the Review page with your selected file.
Click on the thumbnail of the document.
On the left side you will see a preview of the document (document view), and on the right side you will see all the fields with the predictions (data view). Click on the red predictions and check whether they are correct or not. If they are wrong or empty, use your cursor to highlight or draw a box around the correct value in the document view. You can choose your preferred cursor mode in Edit > Cursor mode.
In the data view, when the predictions are in black, it means that the confidence level of that prediction is higher than the set thresholds. In other words, the model is sure that the prediction is correct, thus will not need to be reviewed anymore. If the prediction is in red, the confidence levels of those predictions are lower than the thresholds and thus review is encouraged.
Submit the document once all the information is correct
If you're working in the automated workflow, our system will fetch the next document to review once you submitted the first document. This way you don't have to go back to the inbox and select a new document to review.
This is a quick video of what a 'perfect' document looks like and how you can add a prediction to a new field.
...