Capturing Greater Context for Question Generation

TL;DR

This paper introduces a multi-stage attention-based sequence-to-sequence model that captures broader document context to generate more realistic questions for long documents, outperforming existing methods on multiple datasets.

Contribution

It presents a novel multi-stage attention mechanism to incorporate cross-sentence context in question generation for long documents, improving over prior sentence-level approaches.

Findings

Outperforms state-of-the-art on SQuAD, MS MARCO, NewsQA

Effectively models long document context for question generation

Produces more human-like questions by considering broader context

Abstract

Automatic question generation can benefit many applications ranging from dialogue systems to reading comprehension. While questions are often asked with respect to long documents, there are many challenges with modeling such long documents. Many existing techniques generate questions by effectively looking at one sentence at a time, leading to questions that are easy and not reflective of the human process of question generation. Our goal is to incorporate interactions across multiple sentences to generate realistic questions for long documents. In order to link a broad document context to the target answer, we represent the relevant context via a multi-stage attention mechanism, which forms the foundation of a sequence to sequence model. We outperform state-of-the-art methods on question generation on three question-answering datasets -- SQuAD, MS MARCO and NewsQA.