Clasificador de discurso de odio y discriminatorio - Hate Speech Identification

Example

For an article with the title 'Enough with Syrian kids', the system would likely assign high scores in 'Exaggeration/Attribution/Distortion' due to negative generalization about Syrian children, offering a detailed explanation of the scoring rationale based on specific phrases and suggesting a more neutral headline like 'Addressing challenges faced by Syrian children in local communities'.

Scenario

When a news outlet publishes an article on Venezuelan immigrants that might contain biased language, the classifier can assess the content, provide scores across the hate speech categories, and suggest improvements to mitigate discriminatory tones.

Example

If an article uses hostile language towards LGBTI individuals, the classifier not only assigns scores but also provides a paragraph explaining why the scores were given, citing specific offensive terms used, and recommends a headline that respects gender diversity.

Scenario

In reviewing an op-ed that subtly insults a community, the classifier elucidates the subtleties of the language that could be interpreted as discriminatory, helping editors understand the nuances of hate speech and how to avoid it.