LPI-RIT at LeWiDi-2025: Improving Distributional Predictions via Metadata and Loss Reweighting with DisCo

TL;DR

This paper enhances the DisCo neural architecture for modeling annotator disagreement by incorporating annotator metadata and multi-objective training, leading to improved distributional predictions and evaluation metrics across multiple datasets.

Contribution

We extend DisCo with annotator metadata embeddings and multi-objective losses, significantly improving disagreement modeling in distributional predictions.

Findings

Improved soft and perspectivist evaluation metrics across datasets

Enhanced calibration and error analysis insights

Disagreement modeling benefits from annotator metadata and targeted training

Abstract

The Learning With Disagreements (LeWiDi) 2025 shared task aims to model annotator disagreement through soft label distribution prediction and perspectivist evaluation, which focuses on modeling individual annotators. We adapt DisCo (Distribution from Context), a neural architecture that jointly models item-level and annotator-level label distributions, and present detailed analysis and improvements. In this paper, we extend DisCo by introducing annotator metadata embeddings, enhancing input representations, and multi-objective training losses to capture disagreement patterns better. Through extensive experiments, we demonstrate substantial improvements in both soft and perspectivist evaluation metrics across three datasets. We also conduct in-depth calibration and error analyses that reveal when and why disagreement-aware modeling improves. Our findings show that disagreement can be better captured by conditioning on annotator demographics and by optimizing directly for distributional metrics, yielding consistent improvements across datasets.