Abstract
Although the existing numerical table-to-text generation models have achieved remarkable progress, the idea of generating an accurate analysis of the input table is not well explored. Most existing table-to-text generation algorithms for generating table related information only copy the table record directly but ignore reasoning or calculating over table records. One of the key steps to achieve this ability is number reasoning, which refers to do logical reasoning about the numbers from table records. In this paper, we attempt to improve the number reasoning capability of neural table-to-text generation by generating additional mathematical equations from numerical table records. We propose a neural architecture called Neural Table Reasoning Generator (NTRG), with an additional switching gate as well as a specifically designed equation decoder for generating mathematical equations adaptively. Moreover, we present a pre-training strategy for NTRG similar to the mask language model. Empirical results show that NTRG yields new state-of-the-art results on ROTOWIRE. Furthermore, in order to give a quantitative evaluation of the ability of number reasoning, we construct a sentence-level number reasoning dataset. Results demonstrate the superiority of our approaches over strong baselines.
Similar content being viewed by others
Notes
On publication, we will release our source code and dataset.
In this paper, we only compare each model using the newest evaluation models [22] on ROTOWIRE. Since the test sets of [9] and the evaluation models in [13] are not completely consistent with other works, and some data cannot be obtained, such as the writer information, so no direct comparison is performed.
References
Bai Y, Li Z, Ding N, Shen Y, Zheng HT (2020) Infobox-to-text generation with tree-like planning based attention network. In: Bessiere C (ed) Proceedings of the twenty-ninth international joint conference on artificial intelligence, IJCAI-20, international joint conferences on artificial intelligence organization, pp 3773–3779, main track
Barzilay R, Lapata M (2005) Collective content selection for concept-to-text generation. In: EMNLP, association for computational linguistics, pp 331–338
Devlin J, Chang MW, Lee K, Toutanova K (2018) Bert: pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:181004805
Feng X, Sun Y, Qin B, Gong H, Sun Y, Bi W, Liu X, Liu T (2020) Learning to select bi-aspect information for document-scale text content manipulation. In: AAAI, pp 7716–7723
Gong H, Feng X, Qin B, Liu T (2019) Table-to-text generation via row-aware hierarchical encoder. CCL. Springer, Berlin, pp 533–544
Gong H, Feng X, Qin B, Liu T (2019b) Table-to-text generation with effective hierarchical encoder on three dimensions (row, column and time). In: EMNLP-IJCNLP, pp 3134–3143
Gu J, Lu Z, Li H, Li VO (2016) Incorporating copying mechanism in sequence-to-sequence learning. arXiv preprint arXiv:160306393
Huang D, Shi S, Lin CY, Yin J, Ma WY (2016) How well do computers solve math word problems? Large-scale dataset construction and evaluation. In: ACL (volume 1: long papers), pp 887–896
Iso H, Uehara Y, Ishigaki T, Noji H, Aramaki E, Kobayashi I, Miyao Y, Okazaki N, Takamura H (2019) Learning to select, track, and generate for data-to-text. arXiv preprint arXiv:190709699
Kiddon C, Zettlemoyer L, Choi Y (2016) Globally coherent text generation with neural checklist models. EMNLP 2016:329–339
Li L, Wan X (2018) Point precisely: towards ensuring the precision of data in generated texts using delayed copy mechanism. In: COLING, pp 1044–1055
Li Z, Lin Z, Ding N, Zheng HT, Shen Y (2020) Triple-to-text generation with an anchor-to-prototype framework. In: Bessiere C (ed) Proceedings of the twenty-ninth international joint conference on artificial intelligence, IJCAI-20, international joint conferences on artificial intelligence organization, pp 3780–3786, main track
Lin YC, Yang PA, Lee YK, Chuang KT (2016) Generation of conceptual-level text cloud with graph diffusion. In: Proceedings of the 28th conference on computational linguistics and speech processing (ROCLING 2016), pp 402–411
Liu Q, Guan W, Li S, Kawahara D (2019) Tree-structured decoding for solving math word problems. In: Proceedings of the 2019 conference on empirical methods in natural language processing and the 9th international joint conference on natural language processing (EMNLP-IJCNLP), pp 2370–2379
Locascio N, Narasimhan K, DeLeon E, Kushman N, Barzilay R (2016) Neural generation of regular expressions from natural language with minimal domain knowledge. arXiv preprint arXiv:160803000
Luong MT, Pham H, Manning CD (2015) Effective approaches to attention-based neural machine translation. arXiv preprint arXiv:150804025
Mei H, Bansal M, Walter MR (2015) What to talk about and how? Selective generation using LSTMS with coarse-to-fine alignment. arXiv preprint arXiv:150900838
Mikolov T, Sutskever I, Kai C et al (2013) Distributed representations of words and phrases and their compositionality[J]. Adv Neural Inf Process Syst 26
Nie F, Wang J, Yao JG, Pan R, Lin CY (2018) Operations guided neural networks for high fidelity data-to-text generation. arXiv preprint arXiv:180902735
Pearson K (1901) LIII. On lines and planes of closest fit to systems of points in space. Lond Edinb Dublin Philos Mag J Sci 2(11):559–572
Pennington J, Socher R, Manning C (2014) Glove: global vectors for word representation. In: EMNLP, pp 1532–1543
Puduppully R, Dong L, Lapata M (2018) Data-to-text generation with content selection and planning. arXiv preprint arXiv:180900582
Puduppully R, Dong L, Lapata M (2019) Data-to-text generation with entity modeling. arXiv preprint arXiv:190603221
Reiter E, Dale R (1997) Building applied natural language generation systems. Nat Lang Eng 3(1):57–87
Roy S, Vieira T, Roth D (2015) Reasoning about quantities in natural language. Trans Assoc Comput Linguist 3:1–13
Shen X, Chang E, Su H, Niu C, Klakow D (2020) Neural data-to-text generation via jointly learning the segmentation and correspondence. In: Proceedings of the 58th annual meeting of the association for computational linguistics, association for computational linguistics, Online, pp 7155–7165. https://doi.org/10.18653/v1/2020.acl-main.641
Shi S, Wang Y, Lin CY, Liu X, Rui Y (2015) Automatically solving number word problems by semantic parsing and reasoning. In: EMNLP, pp 1132–1142
Wallace E, Wang Y, Li S, Singh S, Gardner M (2019) Do NLP models know numbers? Probing numeracy in embeddings. arXiv preprint arXiv:190907940
Wang L, Zhang D, Gao L, Song J, Guo L, Shen HT (2018) MathDQN: solving arithmetic word problems via deep reinforcement learning. In: AAAI
Wang Y, Liu X, Shi S (2017) Deep neural solver for math word problems. In: EMNLP, pp 845–854
Wang Z, Wang X, An B, Yu D, Chen C (2020) Towards faithful neural table-to-text generation with content-matching constraints. In: Proceedings of the 58th annual meeting of the association for computational linguistics, association for computational linguistics, Online, pp 1072–1086. https://doi.org/10.18653/v1/2020.acl-main.101,
Wiseman S, Shieber SM, Rush AM (2017) Challenges in data-to-document generation. arXiv preprint arXiv:170708052
Zhao C, Walker M, Chaturvedi S (2020) Bridging the structural gap between encoding and decoding for data-to-text generation. In: Proceedings of the 58th annual meeting of the association for computational linguistics, association for computational linguistics, Online, pp 2481–2491. https://doi.org/10.18653/v1/2020.acl-main.224,
Acknowledgement
We would like to thank the anonymous reviewers and editors for their helpful comments. This work is supported by the National Key R&D Program of China via grant 2020AAA0106502 and National Natural Science Foundation of China (NSFC) via grant 61906053 and Natural Science Foundation of Heilongjiang via grant YQ2019F008.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Feng, X., Gong, H., Chen, Y. et al. Learning number reasoning for numerical table-to-text generation. Int. J. Mach. Learn. & Cyber. 12, 2269–2280 (2021). https://doi.org/10.1007/s13042-021-01305-9
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s13042-021-01305-9