A Text Annotation Tool with Pre-annotation Based on Deep Learning

• In this paper, we introduce an open-source tool, YEDDA, supported by a pre-annotation module based deep learning. EPAD proposes a novel annotation workflow, combining pre-annotation and manual annotation, which improves the efficiency and quality of annotation. The pre-annotation module can effectively reduce the annotation time, and meanwhile improve the precision and recall of annotation. EPAD also contains some of the mechanisms to facilitate the usage of the pre-annotation module. As a collaborative design, EPAD provides administrators with annotation statistics and analysis functions. Experiments showed that EPAD shortened almost 60.0$$\%$$ of the total annotation time, and improved 12.7$$\%$$ of F-measure for annotation quality.

Ämnesord

NATURVETENSKAP  -- Data- och informationsvetenskap (hsv//swe)

NATURAL SCIENCES  -- Computer and Information Sciences (hsv//eng)

Nyckelord

Annotation tool

Deep learning

Pre-annotation

Artificial intelligence

Computer science

Computers

Collaborative design

Manual annotation

Open source tools

Precision and recall

Statistics and analysis

Text annotations

Publikations- och innehållstyp

ref (ämneskategori)

kon (ämneskategori)