# Question Generation by Transformers

arXiv: https://arxiv.org/abs/1909.05017 · HTML: https://mlsanity.com/paper/1909.05017/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 9 Sep 2019.

## Built on by (most cited first, 10 of 10)

- [AraT5: Text-to-Text Transformers for Arabic Language Generation](https://mlsanity.com/paper/2109.12068/) (2021)
- [Multiple-Choice Question Generation: Towards an Automated Assessment Framework](https://mlsanity.com/paper/2209.11830/) (2022)
- [Downstream Transformer Generation of Question-Answer Pairs with Preprocessing and Postprocessing Pipelines](https://mlsanity.com/paper/2205.07387/) (2022)
- [Automated question generation and question answering from Turkish texts](https://mlsanity.com/paper/2111.06476/) (2021)
- [Automating question generation from educational text](https://mlsanity.com/paper/2309.15004/) (2023)
- [Automatic Generation of Multiple-Choice Questions](https://mlsanity.com/paper/2303.14576/) (2023)
- [Semantic-Based Self-Critical Training For Question Generation](https://mlsanity.com/paper/2108.12026/) (2021)
- [Let's Talk! Striking Up Conversations via Conversational Visual Question Generation](https://mlsanity.com/paper/2205.09327/) (2022)
- [Releasing the CRaQAn (Coreference Resolution in Question-Answering): An open-source dataset and dataset creation methodology using instruction-following models](https://mlsanity.com/paper/2311.16338/) (2023)
- [Auto FAQ Generation](https://mlsanity.com/paper/2405.13006/) (2024)

## Abstract (arXiv)

A machine learning model was developed to automatically generate questions from Wikipedia passages using transformers, an attention-based model eschewing the paradigm of existing recurrent neural networks (RNNs). The model was trained on the inverted Stanford Question Answering Dataset (SQuAD), which is a reading comprehension dataset consisting of 100,000+ questions posed by crowdworkers on a set of Wikipedia articles. After training, the question generation model is able to generate simple questions relevant to unseen passages and answers containing an average of 8 words per question. The word error rate (WER) was used as a metric to compare the similarity between SQuAD questions and the model-generated questions. Although the high average WER suggests that the questions generated differ from the original SQuAD questions, the questions generated are mostly grammatically correct and plausible in their own right.