Dependencies & Relations

Annotate NLP relations like coreference, syntax or semantics to label your data. Prodigy’s relations UI is fast and intuitive, and the tool is fully scriptable, so you can build your own automation.

Fast and flexible annotation

Prodigy's workflows for dependency and relation annotation let you quickly link words and phrases together, building a custom graph with any type of syntactic or semantic relation you're interested in. You can use the relations interface for tasks such as syntactic dependency parsing, coreference resolution, information extraction, slot filling and more.

Read more

Try it live and edit dependencies!

This live demo requires JavaScript to be enabled.

Try it live and edit relations and spans!

This live demo requires JavaScript to be enabled.

Label named entities and relations jointly

In lots of tasks, relations hold between phrases, not just single words. The span highlighting mode lets you label and merge spans like named entities by clicking and dragging across them. The relation annotation mode lets you connect tokens and spans and define directional relationships between them.

See examples

Build powerful workflows for custom domain use cases

The hardest thing about relations tasks is that no two problems are exactly alike – there's always some twist that makes it important to add just a little custom logic. That's where Prodigy's scriptability really shines. You can easily add rules to automatically merge phrases or suggest relations, using spaCy's powerful rules engine or your own entirely custom logic.

Read more
This live demo requires JavaScript to be enabled.

Annotation of events in biomedical literature (BioNLP 2011 GENIA Shared Task)

View the documentation