Modeling annotator expertise: Learning when everybody knows a bit of something

TLDR

Supervised learning from multiple labeling sources is an increasingly important problem, yet annotators may not be consistently accurate across the task domain. This paper develops a probabilistic approach that models annotator reliability and expertise that varies with the data they observe. The authors construct a probabilistic model that jointly estimates true labels and annotator expertise across the input space, yielding classification and annotator models. Experiments show that annotator expertise indeed varies in real tasks and that the proposed method outperforms prior multi‑annotator approaches that assume uniform annotator characteristics.

Abstract

Supervised learning from multiple labeling sources is an increasingly important problem in machine learning and data mining. This paper develops a probabilistic approach to this problem when annotators may be unreliable (labels are noisy), but also their expertise varies depending on the data they observe (annotators may have knowledge about different parts of the input space). That is, an annotator may not be consistently accurate (or inaccurate) across the task domain. The presented approach produces classification and annotator models that allow us to provide estimates of the true labels and annotator variable expertise. We provide an analysis of the proposed model under various scenarios and show experimentally that annotator expertise can indeed vary in real tasks and that the presented approach provides clear advantages over previously introduced multi-annotator methods, which only consider general annotator characteristics.