SARC

A fork of NLPrinceton / SARC.

Project Setup

1. Download the dataset.. Create a folder named dataset that is structured like this (don't forget to extract the files):

   dataset/
   ├─ main/
   |   ├─ comments.json
   |   ├─ test-balanced.csv
   |   └─ train-balanced.csv
   └─ pol/
      ├─ comments.json
      ├─ test-balanced.csv
      └─ train-balanced.csv
   

2. Put the dataset folder at this repo's root directory.
3. Still at the repo's root directory, run git submodule update --init. This is one of the dependencies to create bag-of-n-grams (bong).
4. If you want to use word embedding instead of bong, download 1600-dimensional Amazon GloVe embeddings (NOTE: 2.6 GB compressed, 8.7 GB uncompressed). Then put the extracted .txt file inside the dataset folder.

Run the baseline models

Run one of the following commands: ($EMBEDDING is the file of downloaded GloVe embeddings)

'all' dataset

# Bag-of-Words on all:
python eval.py main -l --min_count 5

# Bag-of-Bigrams on all
python eval.py main -n 2 -l --min_count 5

# Embedding on all
python eval.py main -e -l --embedding dataset/amazon_glove1600.txt

'pol' dataset

# Bag-of-Words on pol
python eval.py pol -l

# Bag-of-Bigrams on pol
python eval.py pol -n 2 -l

# Embedding on pol
python eval.py pol -e -l --embedding dataset/amazon_glove1600.txt

Run the proposed models

'pol' dataset

VADER sentiment analysis scores

python turn-level-sentiment.py pol

---

Original Readme

Evaluation code for the Self-Annotated Reddit Corpus (SARC).

• LREC Paper
• SARC Files

Dependencies: NLTK, scikit-learn, text_embedding.

To recreate the all-balanced and pol-balanced results in Table 2 of the paper:

1. download 1600-dimensional Amazon GloVe embeddings (NOTE: 2.4 GB compressed)

2. set the root directory of the SARC dataset at the top of utils.py

3. run the following ($EMBEDDING is the file of downloaded GloVe embeddings)

• Bag-of-Words on all: python SARC/eval.py main -l --min_count 5
• Bag-of-Bigrams on all: python SARC/eval.py main -n 2 -l --min_count 5
• Embedding on all: python SARC/eval.py main -e -l --embedding $EMBEDDING
• Bag-of-Words on pol: python SARC/eval.py pol -l
• Bag-of-Bigrams on pol: python SARC/eval.py pol -n 2 -l
• Embedding on pol: python SARC/eval.py pol -e -l --embedding $EMBEDDING

If you find this code useful please cite the following:

@inproceedings{khodak2018corpus,
  title={A Large Self-Annotated Corpus for Sarcasm},
  author={Khodak, Mikhail and Saunshi, Nikunj and Vodrahalli, Kiran},
  booktitle={Proceedings of the Linguistic Resource and Evaluation Conference (LREC)},
  year={2018}
}