In this thesis I'm using the data set created by Khodak et al. (https://arxiv.org/abs/1704.05579) and by combining it with additional data from Reddit classify the comments as sarcastic or not sarcastic.
Order of the code is: 1 - features, 2 - reddit_download, 3 - user_embeddings, 4 - master_data, 5 - model.
The thesis and its abstract will be uploaded ASAP.