DiSCoMaT: Distantly Supervised Composition Extraction from Tables in Materials Science Articles

Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics Volume 1: Long Papers, pages 13465–13483 July 9-14, 2023 ©2023 Association for Computational Linguistics DISCOMAT: Distantly Supervised Composition Extraction from Tables in Materials Science Articles Tanishq Gupta1, Mohd Zaki2, Devanshi Khatsuriya3, Kausik Hira4, N. M. Anoop Krishnan4,2, Mausam4,3 1Department of Mathematics, 2Department of Civil Engineering 3Department of Computer Science and Engineering, 4Yardi School of Artificial Intelligence Indian Institute of Technology Delhi {tanishqg2406, mohdzaki1995, devanshikhatsuriya18, kausikhira}@gmail.com {krishnan, mausam}@iitd.ac.in Abstract A crucial component in the curation of KB for a scientific domain (e.g., materials science, foods & nutrition, fuels) is information extraction from tables in the domain’s published research articles. To facilitate research in this direction, we define a novel NLP task of extracting com- positions of materials (e.g., glasses) from tables in material science papers. The task involves solving several challenges in concert, such as tables that mention compositions have highly varying structures; text in captions and full pa- per needs to be incorporated along with data in tables; and regular languages for numbers, chemical compounds and composition expres- sions must be integrated into the model. We release a training dataset comprising 4,408 distantly supervised tables, along with 1,475 manually annotated dev and test tables. We also present DISCOMAT, a strong baseline that combines multiple graph neural networks with several task-specific regular expressions, features, and constraints. We show that DIS- COMAT outperforms recent table processing architectures by significant margins. We re- lease our code and data for further research on this challenging IE task from scientific tables. 1 Introduction Advanced knowledge of a science or engineering domain is typically found in domain-specific re- search papers. Information extraction (IE) from scientific articles develops ML methods to auto- matically extract this knowledge for curating large- scale domain-specific KBs (e.g., (Ernst et al., 2015; Hope et al., 2021)). These KBs have a variety of uses: they lead to ease of information access by do- main researchers (Tsatsaronis et al., 2015; Hamon et al., 2017), provide data for developing domain- specific ML models (Nadkarni et al., 2021), and potentially help in accelerating scientific discover- ies (Jain et al., 2013; Venugopal et al., 2021). Significant research exists on IE from text of re- search papers (see Nasar et al. (2018) for a survey), but less attention is given to IE (often, numeric) from tables. Tables may report the performance of algorithms on a dataset, quantitative results of clinical trials, or other important information. Of special interest to us are tables that mention the composition and properties of an entity. Such ta- bles are ubiquitous in various fields such as food and nutrition (tables of food items with nutritional values, see Tables 1-4 in de Holanda Cavalcanti et al. (2021) and Table 2 in Stokvis et al. (2021)), fuels (constituents and calorific values, see Table 2 in Kar et al. (2022) and Beliavskii et al. (2022)), building construction (components and costs, see Table 4 in Aggarwal and Saha (2022)), materi- als (constituents and properties, see …

Content truncated to 500 words