Efficient Online Scalar Annotation with Bounded Support
This addresses the challenge of manual annotation costs for researchers and practitioners in machine learning, though it appears incremental as it builds on existing pairwise ranking methods.
The paper tackles the problem of efficiently collecting scalar annotations for dataset construction and system evaluation, proposing a hybrid method (EASL) that achieves higher correlation with ground truth while significantly improving annotator efficiency.
We describe a novel method for efficiently eliciting scalar annotations for dataset construction and system quality estimation by human judgments. We contrast direct assessment (annotators assign scores to items directly), online pairwise ranking aggregation (scores derive from annotator comparison of items), and a hybrid approach (EASL: Efficient Annotation of Scalar Labels) proposed here. Our proposal leads to increased correlation with ground truth, at far greater annotator efficiency, suggesting this strategy as an improved mechanism for dataset creation and manual system evaluation.