starling-lab / Textual_Annotation_Interface

The interface lets experts annotate textual data to help a model

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Textual_Annotation_Interface

The interface lets experts annotate relations in textual data to build and test a Boost-SRL model.

About:

The software allows users to annotate relations in a training document to train a boosted SRL model (https://starling.utdallas.edu/software/boostsrl/). Test document can be annotated to check the model's accuracy. The predictions of the model (in the results page) shows how correctly (bright green for high, dark green for low) the correct examples were predicted or how well (dark red for low, bright red for high) examples were classified as incorrect. All the previously annotated files are stored and used for the next training.

Usage Instructions:

  • Start Flask server by running "python app.py".
  • Upload the files to be used for training and testing, annotate and see results.
  • Use "python cleanup.py" to delete the train and test files for restarting application.

About

The interface lets experts annotate textual data to help a model

License:GNU General Public License v3.0


Languages

Language:Python 71.7%Language:HTML 16.9%Language:CSS 5.7%Language:JavaScript 5.7%