Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.abbyy.com/llms.txt

Use this file to discover all available pages before exploring further.

A structured document has a fixed layout — fields appear in the same place on every instance. Common examples are questionnaires, application forms, and tax return forms. Some structured documents have a small number of variants with minor field-position differences. For background, see Extract data from documents in Advanced Designer.
IRS Form 1040 (2020) — example of a structured document

​
Where to build the skill

You can create skills for structured documents in either Skill Designer or Advanced Designer; you can only edit them in Advanced Designer.
  • Skill Designer (cloud) — turn on the Fixed-form documents toggle on the skill, then upload and label some blank forms. See Setting up a Document skill for processing structured documents. The resulting skill includes a Forms activity, visible if you later open it in Advanced Designer.
  • Advanced Designer (Windows desktop) — use this when you need to combine the Forms activity with other Vantage activities (the scenarios below).
If you do not enable the Fixed-form documents toggle in Skill Designer, the skill’s processing flow contains only a Fast Learning activity.
Once a skill’s flow includes a Forms activity together with other activities, or contains multiple Forms activities, you can only change skill properties from Vantage — training and deeper edits must happen in Advanced Designer.

​
Pick a scenario

ScenarioWhen to useKey activities
Forms with unstructured elementsForms also include barcodes, stamps, or variable-length tablesForms + Extraction Rules
Mixed forms and unstructured documentsSame data arrives as either forms or freeform textClassify + IF + Forms + Fast Learning

​
Forms with unstructured elements

Some structured documents include unstructured elements such as barcodes or stamps placed anywhere on the page, or have variable-length tables alongside the fixed layout. To process them, pair a Forms activity with an Extraction Rules activity — Forms handles the structured fields, Extraction Rules handles the unstructured elements (for example, barcodes).
1

Create a Document skill

Open Advanced Designer and click Create Document Skill on the start page.
2

Add and train a Forms activity

On the Activities tab, add a Forms activity. In the Activity Editor, on the Blank Form tab, upload one sample blank form per variant (no more than ten variants recommended) and label the fields you want to extract. See labeling guidelines. Click Train Activity.
3

Test the Forms activity

Open the Test Set tab, upload completed test documents, verify that fields are labeled correctly, and click Test Activity. Review the results.
4

Add and configure an Extraction Rules activity

Back on the Activities tab, add an Extraction Rules activity and configure it in the Activity Editor.
5

Test and publish

Click Test Skill Using Selected Documents. If the results are acceptable, publish the skill. Otherwise, adjust the labeling, retrain, and test again.
Tables and repeating groups. Forms activities can extract tables and repeating groups only when the maximum row count or instance count is known in advance and the boundaries are fixed. Label every row that may appear across all form variants — only rows with data are returned in results. Currently, Forms tables support text values only; for columns containing checkboxes or barcodes, use a repeating group instead. If row count is unknown in advance, use a different Vantage technology.

​
Mixed forms and unstructured documents

Some workflows receive the same information on either a printed form or a freeform document — for example, questionnaires returned as both filled forms and unstructured prose. Combine a Forms activity with a Fast Learning (or Extraction Rules) activity, and route documents to the right path with a Classify activity and an IF activity.
1

Create a Document skill

Open Advanced Designer and click Create Document Skill on the start page.
2

Add and train a Forms activity

On the Activities tab, add a Forms activity. In the Activity Editor, on the Blank Form tab, upload a sample blank form and label its fields. See labeling guidelines. Click Train Activity.
3

Test the Forms activity

Open the Test Set tab, upload completed test documents, verify field labeling, and click Test Activity. Review the results.
4

Add and train a Fast Learning activity

On the Activities tab, add a Fast Learning activity and configure and train it in the Activity Editor.
5

Add a Classify activity at the start

Add a Classify activity (Classify By Company or Classify By Text and Image) at the beginning of the flow. In the Activity Editor, create a class for each document variant, assign classes to your training documents, and train the activity.
6

Branch the flow with an IF activity

Add an IF activity, then connect its branches to the Forms and Fast Learning activities so each document type is routed to the correct extraction path.
7

Test and publish

Click Test Skill Using Selected Documents. If the results are acceptable, publish the skill. Otherwise, adjust the labeling and retrain.

Forms activity

Extract fields from documents with fixed layouts.

Extraction Rules activity

Define rule-based extraction for unstructured elements like barcodes.

Fast Learning activity

Refine extraction on freeform documents using manual review feedback.

Classify activities

Branch a flow by document variant before extraction.