Textual adversarial examples crafted with well-designed perturbation can mislead state-of-the-art natural language models. Most previous works on word-level black-box attacks propose different text ...

There are "common phrases that can make you sound judgmental, even when you don't mean to be," says Melody Wilding. Here's ...

The GitHub Action: Profane content filter maintains over 4,900 swear words from nine different languages. For a list of all supported languages and swear words, see the data directory raw ...

Word-finding difficulty (WFD) - frustrating “tip-of-the-tongue” moments when meaning is clear but the word or sound won’t ...