Please wait a minute...
Data Analysis and Knowledge Discovery  2022, Vol. 6 Issue (5): 44-53    DOI: 10.11925/infotech.2096-3467.2021.0857
Current Issue | Archive | Adv Search |
Question Generation Based on Sememe Knowledge and Bidirectional Attention Flow
Duan Jianyong1,2(),Xu Lishan1,Liu Jie1,2,Li Xin1,2,Zhang Jiaming1,Wang Hao1,2
1School of Information, North China University of Technology, Beijing 100144, China
2CNONIX National Standard Application and Promotion Laboratory, Beijing 100144, China
Download: PDF (1029 KB)   HTML ( 18 ) 
Export: BibTeX | EndNote (RIS)      
Abstract  

[Objective] This paper proposes a question generation model based on sememe knowledge and bidirectional attention flow, aiming to improve the semantics of the questions. [Methods] We developed two strategies to enhance semantics: (I) By integrating the external knowledge of sememe in the embedding layer, we captured the semantic knowledge with a smaller granularity than word vectors, and then enhanced the semantic features of the text itself. In addition, we obtained an expanded sememe knowledge base that is more in line with the semantics of the contextual text through the cosine similarity algorithm. It helped us filter out the sememes creating semantic noise in the original knowledge base, and recommended semantically compliant sememe sets for words labeled with non-semantic origins. (II) We enhanced the semantic representation between texts and answers by incorporating a bidirectional attention flow after the encoding layer. [Results] We evaluated our model with the SQuAD1.1 dataset, and the Bleu_1, Bleu_2, Bleu_3, and Blue_4 reached 46.70%, 31.07%, 22.90%, and 17.48%, respectively. The proposed model outperformed the baseline models. [Limitations] With the bidirectional attention flow, the model needs to extract features of paragraph texts and questions, which demands double memory and time to train the model. [Conclusions] Sememe knowledge and bidirectional attention flow could help the proposed model generate higher-quality questions more in line with human language habits.
Key words： Question Generation      Sememe Knowledge      Cosine Similarity      Bidirectional Attention Flow     
Received: 19 August 2021      Published: 01 March 2022
ZTFLH:  TP391  
Fund:National Natural Science Foundation of China(61972003);Humanities and Social Science Foundation of the Ministry of Education(21YJA740052)
Corresponding Authors: Duan Jianyong,ORCID： 0000-0002-2244-3764     E-mail: duanjy@ncut.edu.cn
Service
E-mail this article
Add to my bookshelf
Add to citation manager
E-mail Alert
RSS
Articles by authors
Jianyong Duan
Lishan Xu
Jie Liu
Xin Li
Jiaming Zhang
Hao Wang

Cite this article:

Duan Jianyong, Xu Lishan, Liu Jie, Li Xin, Zhang Jiaming, Wang Hao. Question Generation Based on Sememe Knowledge and Bidirectional Attention Flow. Data Analysis and Knowledge Discovery, 2022, 6(5): 44-53.

URL:

https://manu44.magtech.com.cn/Jwk_infotech_wk3/EN/10.11925/infotech.2096-3467.2021.0857     OR     https://manu44.magtech.com.cn/Jwk_infotech_wk3/EN/Y2022/V6/I5/44

Encoder-Decoder Model
Question Generation Model Based on the Sememe Knowledge and Bidirectional Attention Flow
Example of the Semantic Tree of the Word “cardinal” in HowNet
硬件 参数
处理器 Intel（R）
操作系统 Linux
CPU核数 8核
GPU块数 4块
GPU型号 GeForce GTX 1080 Ti
GPU大小 12GB
Experimental Hardware Configuration Parameters
参数 值
Epoch 50
Max_len 400
学习率 0.2
Batch_size 18
Dropout 0.3
Beam_size 10
Optimal Parameter Settings for Experiments
模型 Bleu_1 Bleu_2 Bleu_3 Bleu_4
Seq2Seq 31.34 13.79 7.36 4.26
NQG++ 43.09 25.96 17.50 12.28
Ass2s - - - 16.20
s2s-map-gsa 45.69 30.25 22.16 16.85
Our model 46.70 31.07 22.90 17.48
Results of Question Generation Model Based on Sememe Knowledge and Bidirectional Attention Flow （%）
集合中义原的个数 Bleu_1 Bleu_2 Bleu_3 Bleu_4
1 46.21 30.60 22.40 17.08
2 46.70 31.07 22.90 17.48
3 46.64 30.89 22.71 17.38
4 46.37 30.65 22.47 17.05
The Influence of the Number of Sememe in the Collection on Performance of the Model （%）
义原作用域 Bleu_1 Bleu_2 Bleu_3 Bleu_4
仅文本 46.62 30.96 22.75 17.33
仅答案 46.42 30.77 22.53 17.31
文本和答案 46.70 31.07 22.90 17.48
The Influence of Sememe Scope on the Model （%）
模型 Bleu_1 Bleu_2 Bleu_3 Bleu_4
s2s-map-gsa 45.69 30.25 22.16 16.85
+ prediction_sememes 46.17 30.64 22.53 17.20
+ FClayer 46.11 30.41 22.26 16.97
Our Model 46.70 31.07 22.90 17.48
Sememe Knowledge and Bidirectional Attention Flow Mechanism Respectively Affect the Model’s Results （%）
Sample Analysis of Problems Generated by Baseline Model
[1] 吴云芳, 张仰森. 问题生成研究综述[J]. 中文信息学报, 2021, 35(7): 1-9.
[1] ( Wu Yunfang, Zhang Yangsen. A Survey of Question Generation[J]. Journal of Chinese Information Processing, 2021, 35(7): 1-9.)
[2] Sutskever I, Vinyals O, Le Q V. Sequence to Sequence Learning with Neural Networks[OL]. arXiv Preprint, arXiv: 1409.3215.
[3] Heilman M, Smith N A. Good Question! Statistical Ranking for Question Generation[C]// Proceedings of the 2010 North American Chapter of the Association for Computational Linguistics Human Language Technologies Conference. 2010: 1-9.
[4] Mannem P, Prasad R, Joshi A. Question Generation from Paragraphs at UPenn: QGSTEC System Description[C]// Proceedings of the 3rd Workshop on Question Generation. 2010: 84-91.
[5] Du X Y, Shao J R, Cardie C. Learning to Ask: Neural Question Generation for Reading Comprehension[C]// Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics. Stroudsburg, PA, USA: Association for Computational Linguistics, 2017: 1342-1352.
[6] Zhou Q, Yang N, Wei F, et al. Neural Question Generation from Text: A Preliminary Study[C]// Proceedings of the 2017 CCF International Conference on Natural Language Processing and Chinese Computing. 2017: 662-671.
[7] Ma X Y, Zhu Q L, Zhou Y L, et al. Improving Question Generation with Sentence-Level Semantic Matching and Answer Position Inferring[J]. Proceedings of the AAAI Conference on A.pngicial Intelligence, 2020, 34(5): 8464-8471.
[8] 谭红叶, 孙秀琴, 闫真. 基于答案及其上下文信息的问题生成模型[J]. 中文信息学报, 2020, 34(5): 74-81.
[8] ( Tan Hongye, Sun Xiuqin, Yan Zhen. Question Generation Model Based on the Answer and Its Contexts[J]. Journal of Chinese Information Processing, 2020, 34(5): 74-81.)
[9] Pennington J, Socher R, Manning C. GloVe: Global Vectors for Word Representation[C]// Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing. Stroudsburg, PA, USA: Association for Computational Linguistics, 2014: 1532-1543.
[10] Niu Y L, Xie R B, Liu Z Y, et al. Improved Word Representation Learning with Sememes[C]// Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics. Stroudsburg, PA, USA: Association for Computational Linguistics, 2017: 2049-2058.
[11] 闫强, 张笑妍, 周思敏. 基于义原相似度的关键词抽取方法[J]. 数据分析与知识发现, 2021, 5(4): 80-89.
[11] ( Yan Qiang, Zhang Xiaoyan, Zhou Simin. Extracting Keywords Based on Sememe Similarity[J]. Data Analysis and Knowledge Discovery, 2021, 5(4): 80-89.)
[12] Seo M, Kembhavi A, Farhadi A, et al. Bidirectional Attention Flow for Machine Comprehension[OL]. arXiv Preprint, arXiv: 1611.01603.
[13] Vaswani A, Shazeer N, Parmar N, et al. Attention is All You Need[OL]. arXiv Preprint, arXiv: 1706.03762.
[14] Luong T, Pham H, Manning C D. Effective Approaches to Attention-Based Neural Machine Translation[C]// Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing. Stroudsburg, PA, USA: Association for Computational Linguistics, 2015: 1412-1421.
[15] Zhao Y, Ni X C, Ding Y Y, et al. Paragraph-Level Neural Question Generation with Maxout Pointer and Gated Self-Attention Networks[C]// Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing. Stroudsburg, PA, USA: Association for Computational Linguistics, 2018: 3901-3910.
[16] Gu J T, Lu Z D, Li H, et al. Incorporating Copying Mechanism in Sequence-to-Sequence Learning[C]// Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics. Stroudsburg, PA, USA: Association for Computational Linguistics, 2016: 1631-1640.
[17] See A, Liu P J, Manning C D. Get to the Point: Summarization with Pointer-Generator Networks[C]// Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics. Stroudsburg, PA, USA: Association for Computational Linguistics, 2017: 1073-1087.
[18] Gulcehre C, Ahn S, Nallapati R, et al. Pointing the Unknown Words[C]// Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics. Stroudsburg, PA, USA: Association for Computational Linguistics, 2016: 140-149.
[19] Kim Y, Lee H, Shin J, et al. Improving Neural Question Generation Using Answer Separation[OL]. arXiv Preprint, arXiv: 1809.02393.
[20] Wang W H, Yang N, Wei F R, et al. Gated Self-Matching Networks for Reading Comprehension and Question Answering[C]// Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics. Stroudsburg, PA, USA: Association for Computational Linguistics, 2017: 189-198.
[1] Wu Dan,Lu Liuxing. Semantic Changes of Queries from Cross-device Searching[J]. 数据分析与知识发现, 2018, 2(8): 69-78.
  Copyright © 2016 Data Analysis and Knowledge Discovery   Tel/Fax:(010)82626611-6626，82624938   E-mail:jishu@mail.las.ac.cn