From this article, you will learn:
- How Rossum can assist you in extracting data from complex tables
- How to switch to the new interface for annotating line items
- How to annotate line items using our new solution
- What are the current limitations of this feature
Please keep in mind that the described feature is in the early stage and is available to a limited number of users for testing and feedback purposes.
For extracting line items, all clients can use our default Line Items Editor, which is explained here.
How Rossum can help you extract data from complex tables
Most documents contain tables and line items. Ideally, the tables would have a clear layout with structured rows and columns. However, this is rarely the case. To address this, we have developed a new solution that efficiently extracts complex line items. Besides handling structured and straightforward cases, this solution allows you to extract nested values without requiring additional add-ons, like our Magic Items extension.
How to switch to the new interface for annotating line items
When this feature is turned on for your account, you will see a new choice in the line items section called “New line items interface (beta)”. To switch between the new option and the regular editor, click the three dots in the line items section and either check or uncheck this option from the list.
We recommend using one approach to annotate a single document. However, if you decide to switch between the new and old interfaces during annotation, you will not encounter any issues. The screen will refresh, and unconfirmed suggestions will disappear, but your annotated values will remain in place.
How to annotate line items using our new solution
Annotating the value
To begin, draw a bounding box around the value you wish to extract from the document. Once the value is selected, a small window with multiple options will appear.
Based on your input, Rossum will automatically suggest values for the next rows that you can accept or modify.
If you do not want to see automatic suggestions, you can turn them off by unchecking the “Automatic suggestion” option.
You can still ask for suggestions by clicking the “Suggest line items” button, which looks like three stars. When you click it, Rossum starts predicting values, and the button changes to a spinning circle to show that it is working on it.
The value you have chosen shows up in the first box of the pop-up window. You can adjust it manually if needed.
Here you can also pick a label for it, which is the column where the value should go.
If everything is correct, you can accept the value (single checkmark icon).
You can also delete the value (bin icon) and start over if needed.
If there are multiple suggestions, you can accept all of them together (double checkmark icon).
You can also add a new row above or below the selected value if you need to include more information.
To remove unnecessary row, just hover over it and click on the bin icon that appears.
The recommended workflow is as follows:
- Click on the first cell of first row in the footer and annotate relevant value
- Annotate other values in the first row
- When the first row is ready, review suggestions in the following rows and adjust or accept them
- Ensure there are no missing values
- Make any needed changes
- To apply the same pattern on the next page of the document, click the “Extract line items on next page” button
If data is automatically predicted, you can still change or adjust it according to your needs. To do this, fix the value that requires changes and then verify the suggestions made based on that. We recommend always adjusting the value in the first row. To do that simply click the value and the pop-up with more options will appear.
Working with the table
To make annotating easier, we have added some extra features to assist you in your daily tasks.
If your document spans multiple pages and you want to apply the changes from the first page to the next one, simply click on the “Extract line items on next page” button.
To quickly delete data from a specific page, column or row, click on its blue label (like “Page 1”, “Description” or “1”), and the data will be chosen automatically. When you point your mouse at the label, the bounding boxes that will be selected will have a blue border.
To choose multiple bounding boxes, you can also hold down the “Shift” key on your keyboard and draw a selection box by clicking the left mouse button and dragging it.
Current limitations of the feature
Currently, the feature is not compatible with the Distributive Webhook, which is another popular extension that distributes extracted values to all line items.
