NaCTeM

Text Mining Tools


brat: annotation visualization and editing

Overview

Intuitive visualization and editing of text annotations is important for communicating the "meaning" of annotations and for reducing the effort of creating new annotations.

brat rapid annotation tool (brat) is a web-based tool for annotation visualization and editing. The tool is freely available and open source (MIT license).

brat supports a rich set of fully configurable annotation primitives:

  • Typed text spans (e.g. entity mention)
  • Binary relations (e.g. coreference)
  • n-ary associations (e.g. events)
  • Attributes/meta-knowledge (e.g. Negation, Speculation, etc.)
  • Free-form text "notes"

These allow the tool to be applied to a wide range of text annotation tasks, including, for example, entity mention annotation, chunking, binary relation annotation, dependency syntax, and structured n-ary event annotation.

8/11/2012: Version 1.3 released

New features in v1.3 include:

For details, please see: http://brat.nlplab.org/new-in-v1.3.html

Annotation visualization

The brat visualization functionality is based on the stav, a visualization tool created by the Tsujii laboratory of the University of Tokyo for the BioNLP Shared Task 2011. The initial focus of the tool was on the visualization of annotations for event extraction, and the visualization is thus originally designed to work also for complex structured annotations.


example visualization of event annotations

The functionality of brat has been extended to support visualization for many other annotation tasks, and brat provides also support features such as text and annotation search and concordancing in addition to visualization.

Online annotation

brat includes annotation capabilities using intuitive mouse-based editing "gestures" familiar from text editors, presentation software, and many other tools.

An annotation for a text span can be created simply by selecting that span with the mouse:

annotating a text span in brat

Annotations can be connected by "dragging" from one annotation to the other:

annotating a coreference relation in brat

brat has been developed in close collaboration with experienced annotators working on mid-to-large-scale annotation efforts (tens of thousands of annotations), and the tool implements a full set of features for annotation support such as automatic validation of annotations against task-specific semantic constraints.

For more information, please see the brat homepage.

Contributors

The primary stav and brat developers are

Acknowledgments

stav and brat development has been supported in part by

NaCTeM is contributing to the development of brat as a collaborative open-source project.

Availability

brat is freely available with full source code under the open-source MIT license.

(The older visualization tool, stav, has been superseded by brat but remains available from the stav repository.)

References

If you use brat in your work, please cite the following paper:

  • Pontus Stenetorp, Sampo Pyysalo, Goran Topić, Tomoko Ohta, Sophia Ananiadou and Jun'ichi Tsujii (2012). brat: a Web-based Tool for NLP-Assisted Text Annotation. In Proceedings of the Demonstrations Session at EACL 2012. (to appear)

Other related studies:

  • Pontus Stenetorp, Goran Topić, Sampo Pyysalo, Tomoko Ohta, Jin-Dong Kim and Jun'ichi Tsujii (2011). BioNLP Shared Task 2011: Supporting Resources. In Proceedings of BioNLP Shared Task 2011 Workshop. (manuscript introducing stav)