This dataset tests whether a model can detect sufficient perspective diversity in analysis.
The task is simple:
Given a scenario and a perspective-diversity claim, predict whether the claim is supported.
Analysis becomes fragile when it relies on one viewpoint, one role, one metric, or one layer of evidence.
Perspective diversity improves stability by including different system positions.
Useful perspectives may include:… See the full description on the dataset page:
https://huggingface.co/datasets/ClarusC64/perspective-diversity-classification-v0.1.
