This page contains code of the Web application and web service based on the neural argument tagger. The figure below illustrates how the sytem looks like: you can enter some text in the input box and detect arguments in it with one of the pre-trained neural models for argument mining.
Front-end code is based on MIT licensed displacy-ent by explosion.ai. We are thankful to Ines Motani for developing this piece of software and making it publicly available under the MIT license.
More specifically, this repository shares code and data related to the following demo paper:
Artem Chernodub, Oleksiy Oliynyk, Philipp Heidenreich, Alexander Bondarenko, Matthias Hagen, Chris Biemann, and Alexander Panchenko (2019): TARGER: Neural Argument Mining at Your Fingertips. In Proceedings of the 57th Annual Meeting of the Association of Computational Linguistics (ACL'2019). Florence, Italy.
If you use the demo or would like to refer to it, please cite the paper mentioned above. You can also use the following BibTex information for citation:
@inproceedings{chernodub2019targer,
title={TARGER: Neural Argument Mining at Your Fingertips},
author={Chernodub, Artem and Oliynyk, Oleksiy and Heidenreich, Philipp and Bondarenko, Alexander and Hagen, Matthias and Biemann, Chris and Panchenko, Alexander},
booktitle={Proceedings of the 57th Annual Meeting of the Association of Computational Linguistics (ACL'2019)},
year={2019},
address={Florence, Italy}
}
Below you will find some instructions on how to run TARGER web application and its API locally (using docker). Alternatively you can just (access it online)[http://ltdemos.informatik.uni-hamburg.de/targer/] though web interface or using API. This web application relies on a (neural tagging library)[http://github.com/achernodub/targer] based on (PyTorch)[https://pytorch.org] framework. You may find this library also useful on its own i.e. for training sequence taggers for argument mining, but also for other tasks such as part of speech tagging or named entity recognition. The library is taking as input CoNLL files is easy to use and has a minimal number of dependencies. Note tht while this library lives in a separate repository from the web applcation but is also part of the TARGER project. So if you are using the library, please also cite the paper mentioned above. For documentation about the tagging library refer to (its repository)[http://github.com/achernodub/targer].
- Download all the pre-trained model files from: http://ltdata1.informatik.uni-hamburg.de/targer/ to the directory
backend/models:
cd backend
wget http://ltdata1.informatik.uni-hamburg.de/targer/COMBO.h5
wget http://ltdata1.informatik.uni-hamburg.de/targer/ES.h5
wget http://ltdata1.informatik.uni-hamburg.de/targer/ES_dep.h5
wget http://ltdata1.informatik.uni-hamburg.de/targer/IBM.h5
wget http://ltdata1.informatik.uni-hamburg.de/targer/WD.h5
wget http://ltdata1.informatik.uni-hamburg.de/targer/WD_dep.h5
wget http://ltdata1.informatik.uni-hamburg.de/targer/model_new_es.hdf5
wget http://ltdata1.informatik.uni-hamburg.de/targer/model_new_wd.hdf5
- Run the demo using Docker as following:
docker-compose upAccess the frontend at http://localhost:6001/ and the backend REST API at http://localhost:6000.