近年来,数学题的自动求解研究逐渐成为焦点,但是当前研究主要侧重于文字应用题求解,对于几何题的自动求解研究还比较少。针对该问题,已经有研究学者提出了基于深度学习方法的几何题求解模型,但是他们的方法不能根据几何题的特点进行设计,没有将知识点信息应用于题目的求解中。受到人类求解几何题的思维方式的启发,该文基于几何题的求解特点设计了一个几何题知识点预测任务,用于预训练文本编码器,然后从预训练后的文本编码器中获得知识点的语义向量表示。随后设计了一种融合知识点语义信息的几何题求解方法①。实验结果表明,基于知识点预训练任务和知识点信息融合方法的模型能将几何题的自动求解准确率提升至66.89%。
Abstract
Research on the automatic solving of mathematical problems has attracted increasing attention recently, with a focus on Math Word Problems. To address geometric problems, some researchers have proposed geometric problems solving models based on deep learning methods, though not capturing the characteristics of geometric problems. Inspired by the human thinking in solving geometric problems, this paper designs a geometric knowledge points prediction task for Pre-training in which text encoder could obtain a semantic vector representation of knowledge points. Then, this paper also proposes a geometric problem solution method which integrates semantic information of knowledge points. The experimental results show that the proposed model can improve the accuracy to 66.89%.
关键词
数学几何题 /
预训练任务 /
自动求解
{{custom_keyword}} /
Key words
math geometric problems /
pre-training tasks /
automatic solving
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] CHEN J, TANG J, QIN J, et al. GeoQA: A geometric question answering benchmark towards multimodal numerical reasoning[C]//Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, 2021: 513-523.
[2] CAO J, XIAO J. An augmented benchmark dataset for geometric question answering through dual parallel text encoding[C]//Proceedings of the 29th International Conference on Computational Linguistics, 2022: 1511-1520.
[3] LU P, GONG R, JIANG S, et al. Inter-gps: Interpretable geometry problem solving with formal language and symbolic reasoning[C]//Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, 2021: 6774-6786.
[4] SHEN J, YIN Y, LI L, et al. Generate & Rank: A multi-task framework for math word problems[C]//Proceedings of the Association for Computational Linguistics, 2021: 2269-2279.
[5] LIU Y, OTT M, GOYAL N, et al. Roberta: A robustly optimized bert pretraining approach[J]. arXiv preprint arXiv: 1907.11692, 2019.
[6] HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780.
[7] WENTSUN W. Basic principles of mechanical theorem proving in elementary geometries[J]. Journal of Automated Reasoning, 1986, 2: 221-252.
[8] WONG W K, HSU S C, WU S H, et al. LIM-G: Learner-initiating instruction model based on cognitive knowledge for geometry word problem comprehension[J]. Computers & Education, 2007, 48(4): 582-601.
[9] SEO M J, HAJISHIRZI H, FARHADI A, et al. Diagram understanding in geometry questions[C]//Proceedings of the 28th AAAI Conference on Artificial Intelligence, 2014: 2831-2838.
[10] SEO M, HAJISHIRZI H, FARHADI A, et al. Solving geometry problems: Combining text and diagram interpretation[C]//Proceedings of the Conference on Empirical Methods in Natural Language Processing, 2015: 1466-1476.
[11] SACHAN M, XING E. Learning to solve geometry problems from natural language demonstrations in textbooks[C]//Proceedings of the 6th Joint Conference on Lexical and Computational Semantics, 2017: 251-261.
[12] SONG H, WANG Y, ZHANG K, et al. BoB: BERT over BERT for training persona-based dialogue models from limited personalized data[C]//Proceedings of the 59th Annual Meeting of the Association for Computional Linguistics and the 11th International Joint Conference on Natural Language Processing, 2021: 167-177.
[13] WU Q, ZHANG Q, FU J, et al. A knowledge aware sequence to tree network for math word problem solving[C]//Proceedings of the Conference on Empirical Methods in Natural Language Processing: 2020: 7137-7146.
[14] 刘清堂,马鑫倩,周洁,等.融合常识库和语法特征的数学应用题题意理解[J].计算机应用,2023,43(02): 356-364.
[15] YU W, WEN Y, ZHENG F, et al. Improving math word problems with pretrained knowledge and hierarchical reasoning[C]//Proceedings of the Conference on Empirical Methods in Natural Language Processing, 2021: 3384-3394.
[16] DEVLIN J, CHANG M W, LEE K, et al. Bert: Pre-training of deep bidirectional transformers for language understanding[C]//Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2019: 4171-4186.
[17] LAN Z, CHEN M, GOODMAN S, et al. Albert: A lite bert for self-supervised learning of language representations[C]//Proceedings of ICLR,2020: 1-17.
[18] LEWIS M, LIU Y, GOYAL N, et al. BART: Denoising sequence-to-sequence pre-training for natural language generation, translation, and comprehension[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, 2020: 7871-7880.
[19] LIANG Z, ZHANG J, WANG L, et al. MWP-BERT: Numeracy augmented pre-training for math word problem solving[G]//Proceedings of the Association for Computational Linguistics, 2022: 997-1009.
[20] HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016: 770-778.
[21] YU Z, YU J, CUI Y, et al. Deep modular coattention networks for visual question answering[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019: 6281-6290.
[22] AMINI A, GABRIEL S, LIN S, et al. MathQA: Towards interpretable math word problem solving with operation-based formalisms[C]//Proceedings of NAACL-HLT, 2019: 2357-2367.
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
基金
国家自然科学基金(62177015);国防科技重点实验室稳定支持经费项目(WDZC20205250410)
{{custom_fund}}
PDF(1965 KB)
