Exploring the Pāli Canon with Machine Learning

Use of Anaconda is recommended.

• Create an Anaconda environment: conda create --name canon python=3
• Activate the environment: conda activate canon
• Install Anaconda packages: conda install nltk tensorflow scikit-learn matplotlib
• Install non-Anaconda packages: pip install bs4

Run python process.py to download the ATI archive (if needed) and process text into data/sentences.py

Run python analyze.py to train and evaluate a word vector model on the processed sentences. This will produce some examples for evaluation and save a tsne.png file that contains the t-SNE plot for the results. Subsequent runs will use saved model data, unless the data/model directory is deleted.