How About Kind of Generating Hedges using End-to-End Neural Models?

TL;DR

This paper explores generating hedging language in conversational tutoring using fine-tuned neural models and reranking, demonstrating feasibility in noisy, real-world dialogue data.

Contribution

It introduces a neural model-based approach for hedge generation with reranking, applied to tutoring data with disfluencies, highlighting challenges in social and task goals.

Findings

Generation is feasible in noisy dialogue environments.

Reranking improves hedge selection accuracy.

Identifies challenges in balancing social and task-oriented communication.

Abstract

Hedging is a strategy for softening the impact of a statement in conversation. In reducing the strength of an expression, it may help to avoid embarrassment (more technically, ``face threat'') to one's listener. For this reason, it is often found in contexts of instruction, such as tutoring. In this work, we develop a model of hedge generation based on i) fine-tuning state-of-the-art language models trained on human-human tutoring data, followed by ii) reranking to select the candidate that best matches the expected hedging strategy within a candidate pool using a hedge classifier. We apply this method to a natural peer-tutoring corpus containing a significant number of disfluencies, repetitions, and repairs. The results show that generation in this noisy environment is feasible with reranking. By conducting an error analysis for both approaches, we reveal the challenges faced by systems attempting to accomplish both social and task-oriented goals in conversation.