All resources View resource Edit resource DataStore Views Edit resource URL File Clear Upload Name Description FROM THE AUTHORS: The file refined_ngram_dict.csv contains a refined lexicon of n-grams. To get this lexicon we took the set of n-grams of length 1-4 that were contained in our labelled data and for each n-gram calculated the proportion of tweets containing it that were considered as hate speech by the human coders. We then manually went through the lexicon to remove irrelevant terms. You can use Markdown formatting here Format Update Resource