Our codes are written in Python 3.7 for our winning participation at the PAN 2020 Profiling Fake News Spreaders on Twitter task

this repo has the following folders:

• 1_ngram_preprocessing: preprocessing scripts for the n-gram based models
• 2_stat_feature_engineering: feature extraction scripts for descriptive statistics based model
• 3_modeling: training scripts for the unique models
• 4_resampling: train and dev set construction for stacking model
• 5_stackingmodel: trining scripts for stacking model
• final software: the final script uploaded to TIRA
• models: the final trained models and vectorizers uploaded to TIRA for testing
• paper: paper describing our approach

Data

training data available at https://zenodo.org/record/4039435#.X6LCj_NKi00

Citation

If you use our code please cite our work.

@InProceedings{lichouri:2020,
  author =              {Jakab Buda and Flora Bolonyai},
  booktitle =           {{CLEF 2020 Labs and Workshops, Notebook Papers}},
  crossref =            {pan:2020},
  editor =              {Linda Cappellato and Carsten Eickhoff and Nicola Ferro and Aur{\'e}lie N{\'e}v{\'e}ol},
  month =               sep,
  publisher =           {CEUR-WS.org},
  title =               {{An Ensemble Model Using N-grams and Statistical Features to Identify Fake News Spreaders on Twitter--Notebook for PAN at CLEF 2020}},
  url =                 {},
  year =                2020
  }

Contribution

This code was developed by Flora Bolonyai and Jakab Buda