Annotation

Athina can help your team annotate and label your datasets.

Contact us to learn more, or book a demo with our team.

​

Concepts & Terminology

Annotations allow you to label your datasets.

• Annotation View Configuration: A configuration that defines the annotation view.
• Annotation View: A view that displays the annotations.
• Annotation Label: A label that can be applied to a datapoint.
• Annotation Entry: A single annotation entry by an annotator.
• Dataset Subset: A subset of a dataset that is used for annotation.
• Annotator: A user who has been assigned to annotate a dataset or subset.

​

How to create an annotation project

To use annotations:

1. Open the Annotations page from the menubar.

2. Create a project

3. Add a dataset to the project.

4. Create or select an annotation view configuration for the project.

5. Assign annotators to the entire dataset or a subset of the dataset.

​

Annotation Labels

Annotation labels are the labels that can be applied to a datapoint.

You can create and manage Annotation Labels from the Labels tab in the Annotations page.

Labels can be of 3 types:

• Categorical: Allows annotator to select from a list of pre-defined options.
• Numeric: Allows annotator to enter a number.
• Comment: Allows annotator to enter a freeform text comment.

​

Annotation View Configuration

Annotation View Configuration defines the annotation view that annotators will see. This includes:

• The fields (columns) from the dataset that will be displayed.
• The layout of the annotation view.
• The labels that annotators will enter for each datapoint.

​

Fields

Fields are the columns from the dataset that will be displayed in the annotation view.

You can configure which fields will be displayed in the annotation view.

You can also configure the following:

• Whether fields are editable by the annotator or not.
• Whether fields are required to be filled out by the annotator or not.
• Whether fields are expanded / collapsed by default.
• Whether to render the field as a text or markdown.

​

Labels

Labels are the labels that annotators will enter for each datapoint.

You can configure the labels that will be displayed in the annotation view.

​

Annotation Metrics / Results

You can view the metrics / results of the annotation project from the Metrics tab in the Annotations page.

You can also view the metrics of all annotation entries associated with a dataset from the Metrics tab in the Datasets page.

​

Demo Video