Annotation relationships

An annotation type that allows you to connect two annotations.

With the annotation relationships feature, you can create and define relationships between Entity annotations in unstructured text. You can then use these annotation relationships to consolidate labeling workflows and potentially reduce the total amount of language models needed.

Asset type

JSON export

Text

See sample

Images

Not yet supported

Video

Not yet supported

Tiled imagery

Not yet supported

How it works

The two most common use cases for Entity annotation relationships are the following:

Coreference resolution

Coreference resolution is the task of finding all words/expressions that refer to a common word in a body of text. This is important for NLP tasks that involve natural language understanding such as document summarization, question answering, and information extraction.

Coreference resolution usually includes nouns, noun phrases, proper nouns, and pronouns. For example, in the phrase “Freckles, the neighbor's cat, jumped over the fence because he was afraid of the dog”, the words “Freckles”, “cat” and “he” are all referring to the same cat and should be associated via a relationship.

Dependency parsing

Dependency parsing is the process of recognizing the grammatical structure of a sentence based on the dependencies between the words.

The process establishes relationships between “head” words and words which modify those heads. For example, to answer the question “Who is the point guard for the LA Lakers in the next game?” your ML model would need to successfully predict the sentence structure to understand that the user wants to know the point guard of the LA Lakers for the next game.

How to create Entity relationships

  1. Create two Entity annotations.
  1. Use alt+click/drag to create a relationship between the Entity annotations.
  1. Additionally, you can define the relationship.

📘

Note that the relationship itself and the definition are considered annotations.


What’s Next

See what relationships will look like in the Label export.

Did this page help you?