Skip to content
ruitedk6 edited this page Jul 14, 2020 · 2 revisions

HUMAN

HUMAN is an annotation server that stands for...

  • Hierarchical: Supports annotation of hierarchical data. This makes it easy to annotate instances (e.g. online comments) together with their context (e.g. the thread of comments a comment was posted in).
  • Universal: Handles both textual data with and without context as well as PDFs and image annotation.
  • Modular: Various question types (labeling questions, multiple-choice, yes-no, setting bounding boxes etc.) that are self-contained and can be arranged in any order needed. This also makes it easy to implement new custom question types and features.
  • ANnotator: Comes with an easy to use GUI interface for your annotators and project manager.

What is an annotation protocol and the state-machine behind HUMAN?

It is what you make out of it! You start with an idea: The tasks you want annotators to perform on your data, and in which order any questions should show up. Maybe you even want to show certain questions only if an annotator choses to answer a previous question in a certain way? No problem. Maybe you want to loop over a subset of questions? All covered. Or you just want annotators to perform a single, well-defined task? Also fine.

Simply start by writing down your annotation protocol (AP) using the defined AP syntax. The HUMAN AP-Parser takes care of converting your AP into the state-machine that will become the back-bone of your HUMAN annotation server.

The generated state-machine is based on XState and guides the annotator through the tasks defined in your AP.

Start with the introduction to define your own AP.

Clone this wiki locally