Franziska Lorenz—
UX Designer

Annotation Editor

Concept for a general purpose web-based annotation tool for a wide range of linguistic annotations

Project Type

Rework / New feature

This project was done during my time at ExB.

The project

The Annotation Editor is one of ExB's internal tools used by Data Curators to create annotation data that serves to train named entity-, relation- and text classification recognizer. Annotations can be used as feedback for the training of recognizer, so that they can be improved over time.
Previously, the annotator was an external Java tool. Since the goal was to integrate it permanently into the Cognitive Workbench, I was asked to record the current status of the Java tool and to create a revision proposal. In the summer of 2018 the annotation editor was moved into the CWB platform.

Screenshot of the previously used Java Annotation Editor for annotating training data for named entity-, relation- and text classification recognizers.

Average reading time: 6 minutes (1332 words)


Based on the existing Java Annotation Editor, my ideas in the first phase and the sketches for the integration into the current platform, we defined a basic set of what needs to be implemented.
The current annotation editor is opened in the Cognitive Workbench from an existing document list. The user interface consists of a header, a list of annotation types, a visual area, and a text area.


Contains general information about the document as well as options like save and close.

Annotation types list

Lists all named entity-, relation-, text classification- and visual types to be annotated by the Data Curators. It also provides the digestive color code and shortcut to annotate. After creation the annotations are listed beneath their types.

Visual area

Includes image or PDF to annotate image information like tables, logos and handwriting.

Textual area

Contains the text of the document whose values can be annotated - for example names, locations, addresses, dates.