Skip to main content

​
Create document fields

For every entity you want to extract, create a corresponding field in the Document Definition.
1

Add a field

In the Document Definition Editor, right-click the document section name and select Create Field.
2

Make it a Text field

Create a Text field.
3

Enable a region

On the General tab, select Can have region.
4

Name the field

In the Name field, enter a name (for example, PreambleSegment). Field names must not contain spaces or non-English characters, and must not start with a number.
Screenshot of the General tab of the field Properties dialog box in the ABBYY FlexiCapture Document Definition Editor, with the field named PreambleSegment and the Can have region and Text segment options selected.
Repeat these steps for each entity. If you use segmentation, create a separate text field for each segment: create a non-repeating field in a repeating group, select Text segment in the field properties, and select Allow multiple regions if a segment can begin and end on different pages.

​
Create a segmentation model

Segmentation improves the accuracy and speed of entity extraction and is optional. It requires a dedicated NLP model, and you can have only one segmentation model per document section.
1

Open the section properties

In the Document Definition Editor, right-click the document section name and select Properties. On the NLP tab, click Create.
2

Name the model and set its type

In the Name field, enter a name (for example, SegmentationModel). Set Model type to Segmentation, and select the required Language.
Screenshot of the New NLP Model dialog box on the NLP tab of the Document Section Properties in ABBYY FlexiCapture, with the model named SegmentationModel, the Source set to Section, the Model type set to Segmentation, and the Language set to English.
3

Choose the segment fields

Click Next, specify the fields the segments will be extracted into, and click OK.
After you create a segmentation model, train it on sample documents.

​
Create an entity extraction model

To extract entities, you need an entity extraction NLP model trained on manually marked-up documents.
1

Open the section properties

In the Document Definition Editor, open the document section properties, click the NLP tab, and click Create.
2

Name the model and set its type

Enter a Name (for example, EntitiesExtraction). For the data source, select a section (if you do not use segmentation) or a segment (if you do). Set Model type to Extraction, and select the required Language.
3

Choose the result fields

Click Next and choose the result fields to extract from the selected section or segment.
Repeat these steps for each document segment or section from which entities should be extracted. Then save your Document Definition (Document Definition → Save), close it (Document Definition → Close), and publish it (Document Definition → Publish). After you create an entity extraction model, train it on sample documents.
The Allow training option lets you train an NLP model during document processing: the model is trained when you train field extraction using a field extraction training batch. To disable or delete training results, right-click the training batch and select Disabled or Delete on the shortcut menu.