Diversify Question Generation with Retrieval-Augmented Style Transfer

TL;DR

This paper introduces RAST, a retrieval-augmented style transfer framework for question generation that leverages external knowledge and reinforcement learning to enhance expression diversity while maintaining answer consistency.

Contribution

The paper presents a novel retrieval-augmented style transfer method with RL training for diverse question generation, addressing limitations of existing internal knowledge-based approaches.

Findings

Outperforms previous diversity-driven baselines in question diversity

Maintains comparable consistency scores with existing methods

Utilizes external retrieval to enhance expression variety

Abstract

Given a textual passage and an answer, humans are able to ask questions with various expressions, but this ability is still challenging for most question generation (QG) systems. Existing solutions mainly focus on the internal knowledge within the given passage or the semantic word space for diverse content planning. These methods, however, have not considered the potential of external knowledge for expression diversity. To bridge this gap, we propose RAST, a framework for Retrieval-Augmented Style Transfer, where the objective is to utilize the style of diverse templates for question generation. For training RAST, we develop a novel Reinforcement Learning (RL) based approach that maximizes a weighted combination of diversity reward and consistency reward. Here, the consistency reward is computed by a Question-Answering (QA) model, whereas the diversity reward measures how much the final output mimics the retrieved template. Experimental results show that our method outperforms previous diversity-driven baselines on diversity while being comparable in terms of consistency scores. Our code is available at https://github.com/gouqi666/RAST.