Skip to content

A treebank annotation tool based on a statistical parser that is re-trained during annotation

License

Notifications You must be signed in to change notification settings

andreasvc/activedop

Repository files navigation

Active DOP treebank annotation tool

A treebank annotation tool based on a statistical parser that is re-trained during annotation. Paper: http://www.aclweb.org/anthology/C18-2009

screenshot of annotation tool

Installation instructions

Running the demo on a toy treebank and annotation task:

  • extract the example grammar: "discodop runexp example.prm" The grammar will be extracted from "treebankExample.mrg", and the annotation task will consist of the sentences in "newsentsExample.txt".
  • run "FLASK_APP=app.py flask initdb"
  • run "FLASK_APP=app.py flask initpriorities"
  • start the web server with "FLASK_APP=app.py flask run --with-threads". open browser at http://localhost:5000/ username "JoeAnnotator", password "example"

Edit "settings.cfg" to use a different grammar and sentences to annotate, and to configure usernames and passwords. Note that the treebank on which the grammar is based needs to be available, in the paths specified in the grammar parameter file.

Sentences need to be segmented, one sentence per line. For best results, tokenize the sentences to annotate according to treebank conventions.

Reference

bibtex:

@InProceedings{vancranenburgh2018active,
    author={van Cranenburgh, Andreas},
    title={Active DOP: A constituency treebank annotation tool with online learning}
    year={2018},
    booktitle={Proceedings of COLING system demonstrations},
    pages={38--42},
    url={http://www.aclweb.org/anthology/C18-2009}
}

About

A treebank annotation tool based on a statistical parser that is re-trained during annotation

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published