Learn & Extract

Introduction

When training or simply extracting documents in Xtracta, users can train line items and apply learning more easily using the Learn & Extract feature.
Users can extract an entire line-item section by selecting the values in one representative row. The AI then uses that pattern to extract the remaining rows and fields.

This tool assumes that the line items are not “complex” line items, although it may still provide some benefit for complex line items.

Learn & Extract

▶

How to Use Learn & Extract

  1. After uploading a file, access the document in the New ELS.
  2. Add a Row to enable Learn & Extract; optionally, remove all existing rows first.
  3. Select field values from the line in the document based on the field mapping.
  4. Once at least one row has been trained, click the Learn & Extract button.
  5. Finish processing the document and correct any fields as required, then Save, output, or reassign the document as appropriate.
Learn & Extract animation
Information Circle

Using the Learn & Extract tool does not trigger learning until the document is output or the user selects Save and Learn.

For the line items to extract correctly, a stable pattern is required since subsequent lines are extracted using the pattern trained from the initial line or lines.

Although only one line is required for Learn and Extract to work, we recommend training two or more for more challenging scenarios.