EmojiRecommender package implements two models for recommending emojis for an incoming tweet.
The first one is the average model and another is the LSTM model

Average Model:
1. Please follow the instructions in (https://github.com/bdhingra/tweet2vec) to set up the environment for modified_tweet2vec
2. Sample input file is in modified_tweet2vec/data/input.csv
3. Run average_main.py in the parent directory within "EmojiRecommender"
4. This will run the preprocessing, tweet2vec encoding, emoji embeddings, cross validation,
  recommendations and prints accuracy for each fold of cross validation
5. The output files for each fold of cross validation are saved in modified_tweet2vec/data/RecommendEmojis1.csv, modified_tweet2vec/data/RecommendEmojis2.csv, etc

LSTM Model:
1. Install Keros and tensor flow backend
2. Run the shell script preprocessing.sh in the parent directory.
3. This will output a preprocessed file input_preprocessed_2.csv in modified_tweet2vec/data/
4. Use this as the input for LSTM.py inside LSTM model directory