Code for the experiments conducted in the context of the Deep Learning & Digital Humanities Seminar. For a detailed analysis of the results please take a look at this term paper.
Clone the repository:
git clone --recurse-submodules git@github.com:DrCracket/dldh-experiments.git
Install requirements:
pip install -r requirements.txt
Put the annotated poetry data in to the poems
folder. The files are expected to be in unix-flavored csv format.
Put the file with real poems in the real_poems
folder. The file should
contain all real poems separated with a newline.
Train models and compute features:
python gppl.py && python crowdgppl.py
Run the experiments:
python main.py
Compute various statistics of the dataset:
python statistics.py