*Said "dishing" - Serving nutritious texts on smaller dishes.
Authors:
Raul Sena, Victor Bursztyn and Ygor Canalli. - {raulsf,vbursztyn,canalli}@cos.ufrj.br
Corpus credits:
CSTNews (by Aleixo, Pardo et al.) - http://www.icmc.usp.br/~taspardo/sucinto/index.html
#1. Requirements:
- Python 2.7 (restriction due to OR-tools dependency)
- MongoDB
- Virtualenv (includes suitable versions for Python and pip):
- Create and activate your virtualenv
- pip install -r requirements.txt
- OR-tools
- In case pip fails at OR-tools, please follow the instructions to install it manually [ https://developers.google.com/optimization/installing ]
- In the worst case, download and easy_install its egg directly [ https://pypi.python.org/simple/ortools/ ]
- NLTk Trainer on MacMorpho
- Activate your virtualenv
- At project's root:
- python -m nltk.downloader punkt mac_morpho stopwords * python $VIRTUAL_ENV/src/nltk-trainer-master/train_tagger.py mac_morpho --filename aggregator/pos_tagger/mac_morpho_pos_tagger.pickle --no-eval
Ensure you have all dependencies correctly installed before attempting to run DESHIN.
#2. Usage:
- Pre-processing only:
- Activate your virtualenv
- bash run_etl.sh
- Batch summarizer only:
- Activate your virtualenv
- bash run_experiment.sh
- Entire workflow:
- Activate your virtualenv
- bash run_all.sh