Question generation is a conditioned language generation task that consists in generating a context-aware question given a context and the targeted answer. Train language modelling with a mere likelihood maximization has been widely used while suffering from exposure bias and the discordance between the training and the test metrics. In the way of addressing this issue, The presented work portrays a fully Transformer-based reinforcement learning generator-evaluation architecture for neural question generation. To edge the flexibility of the generation, a semantic-based reward score was externally infused during the training to drive the training of the language model. The global architecture is laid out in a generator-evaluator fashion optimized directly to n-gram and semantic-based metrics. Evaluation metrics for language modelling only based on n-gram overlapping do not consider semantic relations between reference and candidate sequences. To improve the evaluation step, a two-fold evaluation was carried out. On the one side, an n-gram overlapping evaluation using the BLEU score. On the other side, a semantic-based assessment using BERTScore and NUBIA. The results were corroborated by a binary human evaluation of the semantic relatedness of the generated question and the ground truth. The results obtained showed that use a semantic-based REINFORCE algorithm for the question generation syntactically reshapes the generated questions while preserving their underlying semantic meaning. Many downstream applications can be drawn from a successful question generation including the enlargement of question answering datasets, the improvement of conversational systems, the enhancement of autonomous educational assessment systems, and so forth.
| Published in | American Journal of Information Science and Technology (Volume 5, Issue 4) |
| DOI | 10.11648/j.ajist.20210504.12 |
| Page(s) | 93-97 |
| Creative Commons |
This is an Open Access article, distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution and reproduction in any medium or format, provided the original work is properly cited. |
| Copyright |
Copyright © The Author(s), 2021. Published by Science Publishing Group |
Question Generation, Transformer, Reinforcement Learning, Generator-Evaluator, Language Generation
| [1] | Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. Attention is all you need, 2017. |
| [2] | Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. Bert: Pre-training of deep bidirectional transformers for language understanding, 2019. |
| [3] | Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J. Liu. Exploring the limits of transfer learning with a unified text-to-text transformer, 2020. |
| [4] | Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, and Sergey Zagoruyko. End-to-end object detection with transformers, 2020. |
| [5] | Chulhee Yun, Srinadh Bhojanapalli, Ankit Singh Rawat, Sashank J. Reddi, and Sanjiv Kumar. Are transformers universal approximators of sequence-to-sequence functions?, 2020. |
| [6] | Yu Chen, Lingfei Wu, and Mohammed J. Zaki. Reinforcement learning-based graph-to-sequence model for natural question generation, 2020. |
| [7] | Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. Bleu: a method for automatic evaluation of machine translation. 2002. URL https://www.aclweb.org/anthology/P02-1040.pdf |
| [8] | Kevin Clark, Minh-Thang Luong, Quoc V. Le, and Christopher D. Manning. Electra: Pre-training text encoders as discriminators rather than generators, 2020. |
| [9] | Michael Heilman. Automatic factual question generation from text. 2015. URL https://lti.cs.cmu.edu/ sites/default/files/research/thesis/2011/michael_heilman_automatic_factual_question_ generation_for_reading_assessment.pdf. |
| [10] | Qingyu Zhou, Nan Yang, Furu Wei, Chuanqi Tan, Hangbo Bao, and Ming Zhou. Neural question generation from text: A preliminary study, 2017. |
| [11] | Patrick Lewis, Yuxiang Wu, Linqing Liu, Pasquale Minervini, Heinrich Küttler, Aleksandra Piktus, Pontus Stenetorp, and Sebastian Riedel. Paq: 65 million probably-asked questions and what you can do with them, 2021. |
| [12] | Kettip Kriangchaivech and Artit Wangperawong. Question generation by transformers, 2019. |
| [13] | Alec Radford, Karthik Narasimhan, Tim Salimans, and Ilya Sutskever. Improving language understanding by generative pre-training. 2020. URL https://cdn.openai.com/research-covers/language-unsupervised/language_understanding_paper.pdf. |
| [14] | Steven J. Rennie, Etienne Marcheret, Youssef Mroueh, Jarret Ross, and Vaibhava Goel. Self-critical sequence training for image captioning, 2017. |
| [15] | Sudha Rao and Hal Daumé III. Answer-based adversarial training for generating clarification questions, 2019. |
| [16] | Yao Zhao, Xiaochuan Ni, Yuanyuan Ding, and Qifa Ke. Paragraph-level neural question generation with max out pointer and gated self-attention networks. 2018. URL https://www.aclweb.org/anthology/D18-1424.pdf. |
| [17] | Qingyu Zhou, Nan Yang, Furu Wei, and Ming Zhou. Sequential copying networks, 2018. |
| [18] | Bang Liu, Mingjun Zhao, Di Niu, Kunfeng Lai, Yancheng He, Haojie Wei, and Yu Xu. Learning to generate questions by learning what not to generate. The World Wide Web Conference on - WWW ’19, 2019. doi:10.1145/3308558.3313737. URL http://dx.doi.org/10.1145/3308558.3313737. |
| [19] | Linfeng Song, Yue Zhang, Zhiguo Wang, and Daniel Gildea. A graph-to-sequence model for amr-to-text generation, 2018. |
| [20] | Xingwu Sun, Jing Liu, Yajuan Lyu, Wei He andYanjun Ma, and Shi Wang. Answer-focused and position-aware neural question generation. 2018. URL https://www.aclweb.org/anthology/D18-1427.pdf. |
| [21] | Weizhen Qi, Yu Yan, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang, and Ming Zhou. Prophetnet: Predicting future n-gram for sequence-to-sequence pre-training, 2020. |
| [22] | Pranav Rajpurkar, Jian Zhang, Konstantin Lopyrev, and Percy Liang. Squad: 100,000+ questions for machine comprehension of text, 2016. |
| [23] | Tianyi Zhang, Varsha Kishore, Felix Wu, Kilian Q. Weinberger, and Yoav Artzi. Bertscore: Evaluating text generation with bert, 2020. |
| [24] | Hassan Kane, Muhammed Yusuf Kocyigit, Ali Abdalla, Pelkins Ajanoh, and Mohamed Coulibali. Nubia: Neural-based interchangeability assessor for text generation, 2020. |
APA Style
Loïc Kwate Dassi. (2021). Semantic-Based Self-Critical Training for Question Generation. American Journal of Information Science and Technology, 5(4), 93-97. https://doi.org/10.11648/j.ajist.20210504.12
ACS Style
Loïc Kwate Dassi. Semantic-Based Self-Critical Training for Question Generation. Am. J. Inf. Sci. Technol. 2021, 5(4), 93-97. doi: 10.11648/j.ajist.20210504.12
AMA Style
Loïc Kwate Dassi. Semantic-Based Self-Critical Training for Question Generation. Am J Inf Sci Technol. 2021;5(4):93-97. doi: 10.11648/j.ajist.20210504.12
Share This Article
Twitter
Linked In
Facebook
Copy
Download@article{10.11648/j.ajist.20210504.12,
author = {Loïc Kwate Dassi},
title = {Semantic-Based Self-Critical Training for Question Generation},
journal = {American Journal of Information Science and Technology},
volume = {5},
number = {4},
pages = {93-97},
doi = {10.11648/j.ajist.20210504.12},
url = {https://doi.org/10.11648/j.ajist.20210504.12},
eprint = {https://article.sciencepublishinggroup.com/pdf/10.11648.j.ajist.20210504.12},
abstract = {Question generation is a conditioned language generation task that consists in generating a context-aware question given a context and the targeted answer. Train language modelling with a mere likelihood maximization has been widely used while suffering from exposure bias and the discordance between the training and the test metrics. In the way of addressing this issue, The presented work portrays a fully Transformer-based reinforcement learning generator-evaluation architecture for neural question generation. To edge the flexibility of the generation, a semantic-based reward score was externally infused during the training to drive the training of the language model. The global architecture is laid out in a generator-evaluator fashion optimized directly to n-gram and semantic-based metrics. Evaluation metrics for language modelling only based on n-gram overlapping do not consider semantic relations between reference and candidate sequences. To improve the evaluation step, a two-fold evaluation was carried out. On the one side, an n-gram overlapping evaluation using the BLEU score. On the other side, a semantic-based assessment using BERTScore and NUBIA. The results were corroborated by a binary human evaluation of the semantic relatedness of the generated question and the ground truth. The results obtained showed that use a semantic-based REINFORCE algorithm for the question generation syntactically reshapes the generated questions while preserving their underlying semantic meaning. Many downstream applications can be drawn from a successful question generation including the enlargement of question answering datasets, the improvement of conversational systems, the enhancement of autonomous educational assessment systems, and so forth.},
year = {2021}
}
TY - JOUR T1 - Semantic-Based Self-Critical Training for Question Generation AU - Loïc Kwate Dassi Y1 - 2021/11/17 PY - 2021 N1 - https://doi.org/10.11648/j.ajist.20210504.12 DO - 10.11648/j.ajist.20210504.12 T2 - American Journal of Information Science and Technology JF - American Journal of Information Science and Technology JO - American Journal of Information Science and Technology SP - 93 EP - 97 PB - Science Publishing Group SN - 2640-0588 UR - https://doi.org/10.11648/j.ajist.20210504.12 AB - Question generation is a conditioned language generation task that consists in generating a context-aware question given a context and the targeted answer. Train language modelling with a mere likelihood maximization has been widely used while suffering from exposure bias and the discordance between the training and the test metrics. In the way of addressing this issue, The presented work portrays a fully Transformer-based reinforcement learning generator-evaluation architecture for neural question generation. To edge the flexibility of the generation, a semantic-based reward score was externally infused during the training to drive the training of the language model. The global architecture is laid out in a generator-evaluator fashion optimized directly to n-gram and semantic-based metrics. Evaluation metrics for language modelling only based on n-gram overlapping do not consider semantic relations between reference and candidate sequences. To improve the evaluation step, a two-fold evaluation was carried out. On the one side, an n-gram overlapping evaluation using the BLEU score. On the other side, a semantic-based assessment using BERTScore and NUBIA. The results were corroborated by a binary human evaluation of the semantic relatedness of the generated question and the ground truth. The results obtained showed that use a semantic-based REINFORCE algorithm for the question generation syntactically reshapes the generated questions while preserving their underlying semantic meaning. Many downstream applications can be drawn from a successful question generation including the enlargement of question answering datasets, the improvement of conversational systems, the enhancement of autonomous educational assessment systems, and so forth. VL - 5 IS - 4 ER -
Information
Email: