Topic Modelling Brazilian Supreme Court Lawsuits

This repo holds the source code for the work described in the paper below:

• Pedro H. Luz de Araujo and Teo de Campos
  Topic Modelling Brazilian Supreme Court Lawsuits
  33rd International Conference on Legal Knowledge and Information Systems (JURIX 2020), held online in December 9-11, 2020.

@InProceedings{luz_etal_jurix2020,
      author = {Pedro H. {Luz de Araujo} and Te\'{o}filo E. {de Campos}},
      title = {Topic Modelling Brazilian Supreme Court Lawsuits},
      booktitle = {International Conference on Legal Knowledge and Information Systems ({JURIX})},
      publisher = {IOS Press},
      series = {Frontiers in Artificial Intelligence and Applications},
      pages = {113--122},
      year = {2020},
      month = {December 9-11},
      address = {Prague, Czech Republic},      
      doi = {10.3233/FAIA200855},
      url = {http://ebooks.iospress.nl/volumearticle/56168},
}

The code is available at this repo. The presentation slides are available here.

The sections below describe the requirements and the files.

We kindly request that users cite our paper in any publication that is generated as a result of the use of our work.

Requirements

1. Python 3.6
2. Gensim
3. pyLDAvis
4. Pandas
5. Scikir-learn
6. XGBoost

Files

• lda_explore.ipynb: notebook for topic model training
• train_xgboost_tfidf.ipynb: notebook for classifier training using tfidf values or word counts.
• train_xgboost_topics.ipynb: notebook for classifier training using topic distribution.

Please follow this link to fill in a consent form and get our dataset: http://ailab.unb.br/victor/lrec2020