Author Profiling: Bot and Gender Prediction using a Multi-Aspect Ensemble Approach

This project contains our paper's codes in python, used to indetifying bots and humans and in case of human used to identifying gender in PAN 2019 competition.

Usage

please follow the instructions to profiling authors.

1. running prepare_dataset.py

Parameters:

usage: prepare_dataset.py [-h] [-i INPUT] [-o OUTPUT]

optional arguments:
  -i INPUT      path to input dataset
  -o OUTPUT     path to output directory(default = 'prepared_dataset')

running the script usage:

python prepare_dataset.py -i path_to_dataset_root_dir

2. running training_ngram.py

Parameters:

usage: training_ngram.py [-h] [-i INPUT] [-o OUTPUT] [-ft FT] [-n N]

optional arguments:
  -i INPUT            path to prepared dataset
  -o OUTPUT           path to output directory(default='pre-trained_models')
  -ft FT              frequency threshold (default=5)
  -n N                n-gram order (default=4)

running the script usage:

python training_ngram.py -i path_to_prepared_dataset

3. running training_tfidf.py

Parameters:

usage: training_tfidf.py [-h] [-i INPUT] [-o OUTPUT] [-ft FT]

optional arguments:
  -i INPUT      path to prepared dataset
  -o OUTPUT     path to output directory(default='pre-trained_models')
  -ft FT        frequency threshold (default=5)

running the script usage:

python training_tfidf.py -i path_to_prepared_dataset

4. running training_doc2vec.py

Parameters:

usage: training_doc2vec.py [-h] [-i INPUT] [-o OUTPUT]

optional arguments:
  -i INPUT      path to prepared dataset
  -o OUTPUT     path to output directory(default='pre-trained_models')

running the script usage:

python training_doc2vec.py -i path_to_prepared_dataset

5. running bot_gender_profiling.py

Parameters:

usage: bot_gender_profiling.py [-h] [-i INPUT] [-o OUTPUT] [-t TRAIN_DIR] [-m MODELS] [-n N]

optional arguments:
  -i INPUT        path to dataset directory
  -o OUTPUT       path to output directory
  -t TRAIN_DIR    path to train dataset directory
  -m MODELS       path to models directory
  -n N            n-gram order (default=4)

running the script usage:

python bot_gender_profiling.py -i path_to_dataset_dir -o paht_to_output_dir -t path_to_train_dataset_dir -m paht_to_modles_dir

Citation

Please cite us as:

HB Giglou, M Rahgouy, T Rahgouy, MK Sheykhlan, E Mohammadzadeh. Author Profiling: Bot and Gender Prediction using a Multi-Aspect Ensemble Approach - Notebook for PAN at CLEF 2019. In CLEF 2019 Evaluation Labs and Workshop–Working Notes Papers. CEUR-WS. org.