Enhancing Talk Moves Analysis in Mathematics Tutoring through Classroom Teaching Discourse

TL;DR

This paper introduces SAGA22, a compact dataset for analyzing mathematics tutoring discourse using talk moves, and demonstrates that pretraining on classroom data improves machine learning models' ability to interpret tutoring dialogues.

Contribution

The paper presents a new dataset and modeling strategies that leverage classroom data to enhance talk move analysis in tutoring, addressing scalability challenges.

Findings

Pretraining on classroom data improves model performance.

Longer context and speaker info enhance analysis accuracy.

Ablation studies reveal key challenges in talk move modeling.

Abstract

Human tutoring interventions play a crucial role in supporting student learning, improving academic performance, and promoting personal growth. This paper focuses on analyzing mathematics tutoring discourse using talk moves - a framework of dialogue acts grounded in Accountable Talk theory. However, scaling the collection, annotation, and analysis of extensive tutoring dialogues to develop machine learning models is a challenging and resource-intensive task. To address this, we present SAGA22, a compact dataset, and explore various modeling strategies, including dialogue context, speaker information, pretraining datasets, and further fine-tuning. By leveraging existing datasets and models designed for classroom teaching, our results demonstrate that supplementary pretraining on classroom data enhances model performance in tutoring settings, particularly when incorporating longer context and speaker information. Additionally, we conduct extensive ablation studies to underscore the challenges in talk move modeling.