Question Answering and Question Generation as Dual Tasks

TL;DR

This paper investigates joint question answering and question generation as dual tasks and proposes a differentiable dual-training framework that exploits their probabilistic connection to improve both tasks. It formalizes the QA-QG duality, introduces a joint loss with a duality regularizer, and demonstrates empirical gains across MARCO, SQUAD, and WikiQA, with QA achieving competitive baselines and QG showing BLEU-4 improvements. The core contribution is a from-scratch, end-to-end training approach that leverages probabilistic duality to mutually reinforce QA and QG. This work provides a practical blueprint for applying dual learning to complementary NLP tasks and highlights the potential for cross-task signals to enhance performance in both directions.

Abstract

We study the problem of joint question answering (QA) and question generation (QG) in this paper. Our intuition is that QA and QG have intrinsic connections and these two tasks could improve each other. On one side, the QA model judges whether the generated question of a QG model is relevant to the answer. On the other side, the QG model provides the probability of generating a question given the answer, which is a useful evidence that in turn facilitates QA. In this paper we regard QA and QG as dual tasks. We propose a training framework that trains the models of QA and QG simultaneously, and explicitly leverages their probabilistic correlation to guide the training process of both models. We implement a QG model based on sequence-to-sequence learning, and a QA model based on recurrent neural network. As all the components of the QA and QG models are differentiable, all the parameters involved in these two models could be conventionally learned with back propagation. We conduct experiments on three datasets. Empirical results show that our training framework improves both QA and QG tasks. The improved QA model performs comparably with strong baseline approaches on all three datasets.