A Classification skill determines the type of an incoming document. The document type drives how the document is processed, which fields are extracted, and which rules are checked. It also enables initial manual sorting — for example, routing documents to the correct department.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 Classification skill processes one file per transaction. To classify several files containing documents of different types, use the Classify Activity Process skill.
How it works
To train a Classification skill, specify the classes you want and provide a few example documents for each. Vantage analyzes the text and visual elements of each document — including seals and signatures — so the classifier can handle low-quality images and distinguish between similar document types.When to use a Classification skill
A Classification skill can run as part of a Process skill or on its own:- As part of a Process skill. The classifier sorts each incoming document by type, and the Process skill automatically routes it through the rest of the pipeline.
- On its own. Classify documents directly through the Vantage API, or through a front-end built on it such as the Try Any Skill portal.
Typical use cases
- Sorting mailroom intake into invoices, contracts, and correspondence
- Triaging archived documents before extraction
- Routing inbound requests to the correct department
Relationship to other skills
A Classification skill sits at the front of a document pipeline and decides what type each file is. Other Vantage skills handle what happens next:- Document skill — extracts field data from a document once its type is known.
- OCR skill — produces searchable text from an image or PDF.
- Process skill — orchestrates multiple skills (classification, extraction, validation) into a single workflow.
Common misconceptions
A Classification skill labels documents by type. It does not extract field values, run validation rules, or produce searchable text. For field extraction, pair it with a Document skill. For text recognition only, use an OCR skill.Next steps
Set up a Classification skill
Configure languages, Online learning, and publishing for a new Classification skill.
Train a classifier
Build a training set, assign documents to classes, and run classifier training.
Analyze classifier results
Review per-class accuracy in the Result tab and fix common training-set errors.