Data Annotation

The process of attaching labels, transcriptions, bounding boxes, or other structured metadata to raw data so that it can be used to train, evaluate, or benchmark machine-learning models. Annotation is typically performed by human workers - in-house experts, clinicians, crowdworkers on platforms such as Amazon Mechanical Turk, or community members - and the demographic, cultural, and disability background of these annotators shapes what the resulting model treats as 'correct'. In accessibility contexts, annotation choices can encode or erase disabled people's lived experiences, making annotator selection an ethical as well as a technical decision.