CheapShot is a tool which can detect potentially toxic tweets from any Twitter account.
Firstly, ensure that all the requirements are installed. The CheapShot.py file takes one of three arguments:
| Argument | Shortform | Description |
|---|---|---|
| --user | -u | Search for toxic tweets for single user by username |
| --file | -f | Search for toxic tweets for multiple users by txt file containing list of usernames seperated by a newline |
| --help | -h | Show descriptions for options |
| --limit | -l | Limit the number of tweets collected |
CheapShot uses Twint to source the tweets from a user and then filters them through IBM's toxic comment classifier to discover potentially toxic comments. The toxic comment classifier employs a state-of-the-art pre-trained BERT-Base, English Uncased model to detect potentially toxic comments, it then seperate them into 6 different toxicity types: Toxic, Severe Toxic, Obscene, Threat, Insult and Identity hate.
- MAX-Toxic-Comment-Classifier docker image
- Twint
- Pandas
- BeautifulSoup