Skip to main content
Natural Language Processing (NLP) is a subfield of artificial intelligence and computational linguistics concerned with the computer analysis and synthesis of natural languages. One practical application is extracting meaningful data from text. How a document is processed depends on its structure. There are three types: For more information, see Types of documents processed using ABBYY FlexiCapture. Use NLP to process unstructured documents. For example, NLP can extract reference numbers, names of parties, important dates (signing, effective, term, and termination dates), contract price, fees, and terms of payment from a contract. To extract data from tables or from structured and semi-structured documents, use other methods, such as FlexiLayouts.

​
How NLP extracts information

ABBYY products use NLP models to extract information from unstructured text. An NLP model defines which entities to extract. When you train an NLP model on sample documents, it determines the subject area and the extraction algorithm so the information is extracted more efficiently. The effort needed to create an NLP model depends on the variety of your documents, the context available, and the complexity and volume of information to extract. Extracting data from unstructured text takes significant computing power, and larger texts take longer to analyze. Often, though, the information you need is on a specific page or in a specific paragraph. Finding those useful parts is called segmentation, which takes far less time and fewer resources than entity extraction, so you may want to segment a document before extracting from it. For more information, see Create a segmentation model.

​
Process unstructured documents with NLP

  1. Install the NLP module.
  2. Create a Document Definition.
  3. Create and train an NLP model, or load an existing NLP model into your Document Definition.