######################## # Description of files # ######################## 05 Apr 2019 ./union_svm_bow_vs_combi_annotated.tsv Human judgments on Union dataset ./union_svm_bow_vs_combi.order Since the emojis are randomized, this file provides the indices of [original, bow, combi] emojis respectively. For example [2, 1, 0] means the original emoji is presented as Third Emoji, emoji predicted by BoW system is presented as Second Emoji, and emoji predicted by combi system is presented as First Emoji (and the raters rated accordingly) ./union_train_emoji_emotions.tweet ./union_validation_emoji_emotions.tweet ./union_test_emoji_emotions.tweet The train, validation, and test set, including the predicted emotions ./union_tweet_to_twitter.map The mapping from our internal tweet ID (e.g., all-589001, crawled-153) into Twitter ID (e.g., 812015924374974464, 965773311283118080), because we cannot distribute the tweet texts ./semoji_utils.py The utility script to preprocess the downloaded tweets. See the main() function to see how it can be used to produce the final preprocessed tweets.