Toxicity_detection_context

Toxicity detection w/ and w/o context
Concerning comments existing in a thread.
Context information:
The parent comment.
The discussion topic.
The large dataset is included in the data folder in the form of two CSV files.
gn.csv comprises the out of context annotations.
gc.csv comprises the in-context annotations.
The small dataset will be included soon.

For uploading to OpenML, we considered the label for a binary classification problem and joined the two CSV files into one, gn.csv and gc.csv.

The target variable is label, where:
1 = toxic
0 = non-toxic

Non neccessary columns were removed, such as id, api and parent.

paper_url = "https://arxiv.org/pdf/2006.00998"

original_data_url = "https://github.com/ipavlopoulos/context_toxicity?tab=readme-ov-file"