Measuring Annotator Agreement Generally across Complex Structured, Multi-object, and Free-text Annotation Tasks

TL;DR

This paper evaluates inter-annotator agreement measures for complex annotation tasks, identifying challenges with existing methods and proposing two new, more interpretable measures that improve consistency across diverse annotation types.

Contribution

The paper introduces two novel IAA measures designed for complex annotation tasks, addressing interpretability issues and enhancing consistency across various data types.

Findings

Existing IAA measures struggle with complex tasks

Proposed measures are more interpretable and consistent

Evaluation across seven diverse annotation tasks

Abstract

When annotators label data, a key metric for quality assurance is inter-annotator agreement (IAA): the extent to which annotators agree on their labels. Though many IAA measures exist for simple categorical and ordinal labeling tasks, relatively little work has considered more complex labeling tasks, such as structured, multi-object, and free-text annotations. Krippendorff's alpha, best known for use with simpler labeling tasks, does have a distance-based formulation with broader applicability, but little work has studied its efficacy and consistency across complex annotation tasks. We investigate the design and evaluation of IAA measures for complex annotation tasks, with evaluation spanning seven diverse tasks: image bounding boxes, image keypoints, text sequence tagging, ranked lists, free text translations, numeric vectors, and syntax trees. We identify the difficulty of interpretability and the complexity of choosing a distance function as key obstacles in applying Krippendorff's alpha generally across these tasks. We propose two novel, more interpretable measures, showing they yield more consistent IAA measures across tasks and annotation distance functions.