kurt-mueller-osumc/ner-annotator

Named Entity Recognition (NER) Annotation tool for SpaCy. Generates Traning Data as a JSON which can be readily used.

NER Annotator for Spacy

NER Annotator for SpaCy allows you to create training data for creating a custom NER Model with custom tags.

• Web Application: https://tecoholic.github.io/ner-annotator/
• Linux: Debian Package or an AppImage
• Windows: x64 MSI Installer

Features

• Easy to use UI for labelling entities in text
• Custom labels with color-coding
• Word-level and character-level annotation
• Keyboard shortcuts
• Import existing annotations for review
• Light and dark themes

Screenshots

FAQ

How do I import the annotations to spaCy?

A guide is available here.

I have multiple exported annotations. How do I combined them into one file?

A guide is available here.

What tagging format does the tool use?

The JSON file created by the tool is generic and simply lists the start and end of each entity. As such, the annotations are independent of the tagging format you intend to use (IO, IOB, IOBES, etc.).

Development

Requirements

1. Node JS 14.x
2. Yarn Package Manager
3. Rust (for building desktop versions)

Running it locally for development

1. Open another terminal and start the server for the UI

yarn
yarn serve

Now go to http://localhost:8081/ner-annotator/

Developing the desktop application

The desktop applications have been created using Tauri.

yarn tauri:serve

To build the final binaries run

yarn tauri:build

Credits

1. App Icon - Ornithology icons created by Freepik - Flaticon