Intersectionality¶
Intersectionality is a conceptual framework introduced by Kimberlé Crenshaw (1989) recognizing that people hold multiple, overlapping identity categories (race, gender, sexual orientation, class, disability, etc.), and the interaction of these categories creates qualitatively distinct forms of discrimination and oppression—not merely the sum of individual discriminations.
In hate speech and content moderation, intersectionality highlights that hateful speech targeting individuals at the intersection of multiple identities (e.g., racist and sexist attacks against women of color) differs meaningfully from attacks targeting a single axis. Annotation schemes and detection systems that ignore intersectionality risk missing important patterns.
Key papers¶
- Are You a Racist or Am I Seeing Things? Annotator Influence on Hate Speech Detection on Twitter — Introduces intersectional annotation with a "both racism and sexism" label, recognizing that compound oppression is qualitatively distinct
Related topics¶
- Hate speech detection — domain where intersectionality matters for annotation and fairness
- Social Bias — broader framework for demographic biases in ML systems