README.md

Encoder classes

1. classes_two.npy - contains two classes -- "toxic" , "non-toxic"
2. classes.npy- contains three classes -- "hatespeech" , "normal", "offensive"

Code to load the encoder

from sklearn.preprocessing import LabelEncoder
encoder = LabelEncoder()
encoder.classes_ = np.load(<name_of_the_file>,allow_pickle=True)

Dataset Format

Sample Entry:

{
  "24198545_gab": {
    "post_id": "24198545_gab",
    "annotators": [
      {
        "label": "hatespeech",
        "annotator_id": 4,
        "target": ["African"]
      },
      {
        "label": "hatespeech",
        "annotator_id": 3,
        "target": ["African"]
      },
      {
        "label": "offensive",
        "annotator_id": 5,
        "target": ["African"]
      }
    ],
    "rationales":[
    [0,0,0,0,0,0,0,0,1,0,0,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0],
    [0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],
    [0,0,0,0,0,0,0,0,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0]
    ],
    "post_tokens": ["and","this","is","why","i","end","up","with","nigger","trainee","doctors","who","can","not","speak","properly","lack","basic","knowledge","of","biology","it","truly","scary","if","the","public","only","knew"]
  }
}

🔹post_id : Unique id for each post
🔹annotators : The list of annotations from each annotator
🔹annotators[label] : The label assigned by the annotator to this post. Possible values: [Hatespeech, Offensive, Normal]
🔹annotators[annotator_id] : The unique Id assigned to each annotator
🔹annotators[target] : A list of target community present in the post
🔹rationales : A list of rationales selected by annotators. Each rationales represents a list with values 0 or 1. A value of 1 means that the token is part of the rationale selected by the annotator. To get the particular token, we can use the same index position in "post_tokens"
🔹post_tokens : The list of tokens representing the post which was annotated

Post ids divisions

Post_id_divisions has a dictionary having train, valid and test post ids that are used to divide the dataset into train, val and test set in the ratio of 8:1:1.

Word2Vec Model

We use Common Crawl (840B tokens, 2.2M vocab, cased, 300d vectors, 2.03 GB download) pretrained word vectors from glove repo. This is required only when you plan to run the non-bert deep learning model (cnn-gru, birnn, birnn-scrat). One click download

1. Extract the glove.840B.300d.txt in this folder (Data/)
2. Run this python file to convert the glove model into gensim model.

🟢🟢 You are ready to roll!!! 🟢🟢

Note:

The data uses the label "homosexual" as defined at collection time; other sexual and gender orientation categories have been pruned from the data due to low incidence; the published version of the paper wrongly mentions the LGBTQ category.