Damage Localization and Severity Assessment of a Cable-Stayed Bridge Using a Message Passing Neural Network
Open accessThe DOI matches its Crossref record; 54 of the 54 references it cites were found in a registry or an index.
Abstract
Cable-stayed bridges are damaged by multiple factors such as natural disasters, weather, and vehicle load. In particular, if the stayed cable, which is an essential and weak component of the cable-stayed bridge, is damaged, it may adversely affect the adjacent cables and worsen the bridge structure condition. Therefore, we must accurately determine the condition of the cable with a technology-based evaluation strategy. In this paper, we propose a deep learning model that allows us to locate the damaged cable and estimate its cross-sectional area. To obtain the data required for the deep learning training, we use the tension data of the reduced area cable, which are simulated in the Practical Advanced Analysis Program (PAAP), a robust structural analysis program. We represent the sensor data of the damaged cable-stayed bridge as a graph composed of vertices and edges using tension and spatial information of the sensors. We apply the sensor geometry by mapping the tension data to the graph vertices and the connection relationship between sensors to the graph edges. We employ a Graph Neural Network (GNN) to use the graph representation of the sensor data directly. GNN, which has been actively studied recently, can treat graph-structured data with the most advanced performance. We train the GNN framework, the Message Passing Neural Network (MPNN), to perform two tasks to identify damaged cables and estimate the cable areas. We adopt a multi-task learning method for more efficient optimization. We show that the proposed technique achieves high performance with the cable-stayed bridge data generated from PAAP.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Figures








Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source says what the claim attributes to it.
In the citing paper
“Son et al. [196] proposed that a Graph Neural Network (GNN) can learn graph structure data, which overcame the limitation that a CNN can only learn grid structure data.”
From Deep Learning for Structural Health Monitoring: Data, Algorithms, Applications, Challenges, and Trends (2023) · cites this paper as [196]
The cited source says what the claim attributes to it.
In the citing paper
“Finally, using graph sensor data, the GNN was trained to successfully detect cable damage on cable-stayed bridges.”
From Deep Learning for Structural Health Monitoring: Data, Algorithms, Applications, Challenges, and Trends (2023) · cites this paper as [196]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Adam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946 · cited 1×
“We utilize the ADAM optimizer [55] and train the MPNN model to minimize the loss function, which is defined in Equation (28).”
MethodLearning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation
Cho, K., van Merrienboer, B., Gulcehre, C., Bahdanau, D., Bougares, F., Schwenk, H., & Bengio, Y. (2014) · Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.3115/v1/d14-1179 · Cited by 25,545 · cited 1×
“Typical RNN structures, Gated Recurrent Units (GRUs) [44] and Long Short-Term Memory (LSTM), support the gating of the hidden state and control information flow.”
BackgroundCaruana, R. (1997) · Machine Learning · doi:10.1023/a:1007379606734 · Cited by 6,636 · cited 1×
“The multi-task learning benefits while learning related tasks together [38].”
BackgroundShuman, D. I., Narang, S. K., Frossard, P., Ortega, A., & Vandergheynst, P. (2013) · IEEE Signal Processing Magazine · doi:10.1109/msp.2012.2235192 · Cited by 4,421 · cited 1×
“The edge feature euv is defined as the thresholded Gaussian kernel [52] as presented in the equation below using the XYZ coordinates of the nodes on the girder connected to the 10 cables.”
Method(2011) · Cited by 3,939
Graph Convolutional Neural Networks for Web-Scale Recommender Systems
Ying, R., He, R., Chen, K., Eksombatchai, P., Hamilton, W. L., & Leskovec, J. (2018) · Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining · doi:10.1145/3219819.3219890 · Cited by 3,733 · cited 1×
“Thanks to the rapid development of GNN, which is capable of recent graph prediction, the utilization of deep learning increases in various domains such as traffic forecasting [30,31], recommendation system [32,33], molecular property prediction [34], and natural language processing [35,36].”
BackgroundSpatio-Temporal Graph Convolutional Networks: A Deep Learning Framework for Traffic Forecasting
Yu, B., Yin, H., & Zhu, Z. (2018) · Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2018/505 · Cited by 3,728 · cited 1×
“Thanks to the rapid development of GNN, which is capable of recent graph prediction, the utilization of deep learning increases in various domains such as traffic forecasting [30,31], recommendation system [32,33], molecular property prediction [34], and natural language processing [35,36].”
BackgroundNeural Message Passing for Quantum Chemistry
Gilmer, J., Schoenholz, S. S., Riley, P. F., Vinyals, O., & Dahl, G. E. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1704.01212 · Cited by 3,440 · cited 3×
“MPNN [34] is a general framework of GNN.”
BackgroundDeep Learning‐Based Crack Damage Detection Using Convolutional Neural Networks
Cha, Y., Choi, W., & Büyüköztürk, O. (2017) · Computer-Aided Civil and Infrastructure Engineering · doi:10.1111/mice.12263 · Cited by 3,295 · cited 1×
“CNNs trained with structure images successfully classify surface damage such as concrete cracks and spalling conditions [13–16].”
BackgroundAbdeljaber, O., Avci, O., Kiranyaz, S., Gabbouj, M., & Inman, D. J. (2017) · Journal of Sound and Vibration · doi:10.1016/j.jsv.2016.10.043 · Cited by 1,278 · cited 1×
“Abdeljaber et al. [11] proposed a one-dimensional CNN that extracts features from raw accelerometer signals and classifies damage.”
BackgroundA review on the application of deep learning in system health management
Khan, S., & Yairi, T. (2018) · Mechanical Systems and Signal Processing · doi:10.1016/j.ymssp.2017.11.024 · Cited by 1,133 · cited 1×
“With the advancement of the device fabrication process, artificial intelligence meets the need for fast and accurate problem solving using vast amounts of data collected from sensor devices [5].”
MethodEncoding Sentences with Graph Convolutional Networks for Semantic Role Labeling
Marcheggiani, D., & Titov, I. (2017) · Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/d17-1159 · Cited by 907 · cited 1×
“Thanks to the rapid development of GNN, which is capable of recent graph prediction, the utilization of deep learning increases in various domains such as traffic forecasting [30,31], recommendation system [32,33], molecular property prediction [34], and natural language processing [35,36].”
BackgroundAzimi, M., Eslamlou, A., & Pekcan, G. (2020) · Sensors · doi:10.3390/s20102778 · Cited by 769 · cited 1×
“With the development of computer vision, 2D CNN has been successfully used as a vision-based SHM technique [12].”
MethodDeep Transfer Learning for Image‐Based Structural Damage Recognition
Gao, Y., & Mosalam, K. M. (2018) · Computer-Aided Civil and Infrastructure Engineering · doi:10.1111/mice.12363 · Cited by 766
Sun, L., Shang, Z., Xia, Y., Bhowmick, S., & Nagarajaiah, S. (2020) · Journal of Structural Engineering · doi:10.1061/(asce)st.1943-541x.0002535 · Cited by 731 · cited 15×
“In this study, we assess the GNN-based SHM technology assuming that only one cable is damaged.”
BackgroundInfluentialDiffusion Convolutional Recurrent Neural Network: Data-Driven Traffic Forecasting
Li, Y., Yu, R., Shahabi, C., & Liu, Y. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1707.01926 · Cited by 507
Structural damage identification based on autoencoder neural networks and deep learning
Pathirage, C. S. N., Li, J., Li, L., Hao, H., Liu, W., & Ni, P. (2018) · Engineering Structures · doi:10.1016/j.engstruct.2018.05.109 · Cited by 415 · cited 1×
“[7] trained the autoencoder neural network to perform dimensionality reduction and estimated the stiffness element of the steel frame structure with modal information.”
MethodLarge-Scale Hierarchical Text Classification with Recursively Regularized Deep Graph-CNN
Peng, H., Li, J., He, Y., Liu, Y., Bao, M., Wang, L., Song, Y., & Yang, Q. (2018) · Proceedings of the 2018 World Wide Web Conference on World Wide Web - WWW '18 · doi:10.1145/3178876.3186005 · Cited by 415 · cited 1×
“Thanks to the rapid development of GNN, which is capable of recent graph prediction, the utilization of deep learning increases in various domains such as traffic forecasting [30,31], recommendation system [32,33], molecular property prediction [34], and natural language processing [35,36].”
BackgroundSolution method for nonlinear problems with multiple critical points
Yang, Y.-B., & Shieh, M.-S. (1990) · AIAA Journal · doi:10.2514/3.10529 · Cited by 337 · cited 1×
“The details of the GDC are presented in [17,39].”
BackgroundTerrestrial Laser Scanning-Based Structural Damage Assessment
Olsen, M. J., Kuester, F., Chang, B. J., & Hutchinson, T. C. (2010) · Journal of Computing in Civil Engineering · doi:10.1061/(asce)cp.1943-5487.0000028 · Cited by 311
Rumpf, H. (1990) · Springer eBooks · doi:10.1007/978-94-011-7944-7 · Cited by 311
Image‐based concrete crack assessment using mask and region‐based convolutional neural network
Kim, B., & Cho, S. (2019) · Structural Control and Health Monitoring · doi:10.1002/stc.2381 · Cited by 283
A System for Massively Parallel Hyperparameter Tuning
Li, L., Jamieson, K., Rostamizadeh, A., Gonina, E., Ben-tzur, J., Hardt, M., Recht, B., & Talwalkar, A. (2020) · Proceedings of Machine Learning and Systems · Cited by 236
Geometric Matrix Completion with Recurrent Multi-Graph Neural Networks
Monti, F., Bronstein, M. M., & Bresson, X. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1704.06803 · Cited by 235
Structural health monitoring methods of cables in cable-stayed bridge: A review
Zhang, L., Qiu, G., & Chen, Z. (2021) · Measurement · doi:10.1016/j.measurement.2020.108343 · Cited by 235 · cited 2×
“The quantity of surveyed cables depends on the scale and complexity of the bridge and the monitoring objectives [47,50].”
BackgroundNonlinear static and dynamic analysis of cable structures
Thai, H.-T., & Kim, S.-E. (2011) · Finite Elements in Analysis and Design · doi:10.1016/j.finel.2010.10.005 · Cited by 214 · cited 3×
“The cable element is supposed to be perfectly flexible [40] with the self-weight distributed along its length.”
BackgroundMethodSelf-citationConvolutional neural networks for automated damage recognition and damage type identification
Modarres, C., Astorga, N., Droguett, E. L., & Meruane, V. (2018) · Structural Control and Health Monitoring · doi:10.1002/stc.2230 · Cited by 185
Li, S., Li, H., Liu, Y., Lan, C., Zhou, W., & Ou, J. (2014) · Structural Control and Health Monitoring · doi:10.1002/stc.1559 · Cited by 180
Chang, C.-M., Lin, T.-K., & Chang, C.-W. (2018) · Measurement · doi:10.1016/j.measurement.2018.07.051 · Cited by 148 · cited 1×
“They simulated damaged structures by reducing the elastic modulus of individual elements and verified the performance of the proposed DFNN. Changa et al. [10] estimated damage locations and severity by training neural networks with modal properties after reducing stiffness to create damage patterns.”
MethodFarahani, B. V., Barros, F., Sousa, P. J. S. C. P., Cacciari, P. P., Tavares, P. J., Futai, M. M., & Moreira, P. M. G. P. (2019) · Tunnelling and Underground Space Technology · doi:10.1016/j.tust.2019.102995 · Cited by 115
Damage detection under varying temperature using artificial neural networks
Gu, J., Gul, M., & Wu, X. (2017) · Structural Control and Health Monitoring · doi:10.1002/stc.1998 · Cited by 111
An efficient image-based damage detection for cable surface in cable-stayed bridges
Ho, H.-N., Kim, K.-D., Park, Y.-S., & Lee, J.-J. (2013) · NDT & E International · doi:10.1016/j.ndteint.2013.04.006 · Cited by 105 · cited 1×
“Thus, the potential damage of cables should be identified early to prevent terrible disasters [48,49].”
BackgroundTruong, T. T., Dinh-Cong, D., Lee, J., & Nguyen-Thoi, T. (2020) · Journal of Building Engineering · doi:10.1016/j.jobe.2020.101244 · Cited by 102 · cited 1×
“[9] introduced deep feedforward neural networks (DFNN) to detect damage to truss structures.”
BackgroundMultivariate statistical analysis for early damage detection
Santos, J. P., Crémona, C., Orcesi, A., & Silveira, P. (2013) · Engineering Structures · doi:10.1016/j.engstruct.2013.05.022 · Cited by 93 · cited 1×
“Researchers have introduced deep learning models as well as statistical analysis and machine learning as SHM techniques to determine the damaged cable locations [2] or detect stiffness reduction [4].”
MethodNazarian, E., Ansari, F., Zhang, X., & Taylor, T. (2016) · Journal of Structural Engineering · doi:10.1061/(asce)st.1943-541x.0001463 · Cited by 90 · cited 1×
“Cable Damage Model During the service life of cable-stayed bridges, cables are the most critical load-bearing components [46,47].”
BackgroundVibration‐based semantic damage segmentation for large‐scale structural health monitoring
Sajedi, S. O., & Liang, X. (2020) · Computer-Aided Civil and Infrastructure Engineering · doi:10.1111/mice.12523 · Cited by 89 · cited 1×
“Studies using point clouds produced by laser scanning have been proposed to evaluate the structure state regarding the entire bridge structure and 3D spatial information data between sensors [24–26].”
MethodLarge deflection inelastic analysis of space trusses using generalized displacement control method
Thai, H., & Kim, S.-E. (2009) · Journal of Constructional Steel Research · doi:10.1016/j.jcsr.2009.06.012 · Cited by 83 · cited 1×
“Using the GDC method [20,51] to solve nonlinear problems, the PAAP divides the dead load into many incremental steps.”
MethodSelf-citationSong, M., Yousefianmoghadam, S., Mohammadi, M. E., Moaveni, B., Stavridis, A., & Wood, R. L. (2017) · Structural Health Monitoring · doi:10.1177/1475921717737970 · Cited by 66 · cited 1×
“However, using point cloud data for structural health evaluation is possible only for visible elements such as cracks and spalling [27,28], and the point cloud data are not directly related to the loss of stiffness or strength since the point cloud data do not adequately capture the depth due to occlusion [29].”
BackgroundPractical advanced analysis software for nonlinear inelastic analysis of space steel structures
Thai, H., & Kim, S.-E. (2009) · Advances in Engineering Software · doi:10.1016/j.advengsoft.2009.02.001 · Cited by 63 · cited 3×
“The advantage of the PAAP is that the nonlinear structural responses are accurately obtained with only one or two elements per structural member leading to low computational costs [17,21].”
BackgroundMethodSelf-citationSecond-order plastic-hinge analysis of space semi-rigid steel frames
Ngo-Huu, C., Nguyen, P., & Kim, S.-E. (2012) · Thin-Walled Structures · doi:10.1016/j.tws.2012.06.019 · Cited by 47
Self-citationLi, S., Niu, J., & Li, Z. (2021) · Structural Health Monitoring · doi:10.1177/1475921720988666 · Cited by 41
Nonlinear inelastic time-history analysis of three-dimensional semi-rigid steel frames
Nguyen, P., & Kim, S.-E. (2014) · Journal of Constructional Steel Research · doi:10.1016/j.jcsr.2014.05.009 · Cited by 41
Self-citationTruong, V.-H., & Kim, S.-E. (2017) · Structural and Multidisciplinary Optimization · doi:10.1007/s00158-017-1667-7 · Cited by 39 · cited 2×
“In addition, the reliability of PAAP has been evaluated for the cable-stayed bridges [20,21] and suspension bridges [22,23].”
BackgroundMethodSelf-citationSurface damage quantification of postearthquake building based on terrestrial laser scan data
Dai, K., Li, A., Zhang, H., Chen, S.-E., & Pan, Y. H. (2018) · Structural Control and Health Monitoring · doi:10.1002/stc.2210 · Cited by 33
Improved refined plastic hinge analysis accounting for strain reversal
Kim, S.-E., Kim, M. K., & Chen, W.-F. (2000) · Engineering Structures · doi:10.1016/s0141-0296(98)00079-0 · Cited by 28
Self-citationSecond-order inelastic analysis of cable-stayed bridges
Thai, H., & Kim, S.-E. (2011) · Finite Elements in Analysis and Design · doi:10.1016/j.finel.2011.07.002 · Cited by 27 · cited 3×
“In addition, the reliability of PAAP has been evaluated for the cable-stayed bridges [20,21] and suspension bridges [22,23].”
BackgroundSelf-citationNonlinear inelastic dynamic analysis of suspension bridges
Kim, S.-E., & Thai, H. (2010) · Engineering Structures · doi:10.1016/j.engstruct.2010.08.027 · Cited by 27 · cited 1×
“In addition, the reliability of PAAP has been evaluated for the cable-stayed bridges [20,21] and suspension bridges [22,23].”
BackgroundSelf-citationSuchocki, C., Jagoda, M., Obuchovski, R., Šlikas, D., & Visockienė, J. S. (2018) · Metrology and Measurement Systems · doi:10.24425/mms.2018.124886 · Cited by 22 · cited 1×
“However, using point cloud data for structural health evaluation is possible only for visible elements such as cracks and spalling [27,28], and the point cloud data are not directly related to the loss of stiffness or strength since the point cloud data do not adequately capture the depth due to…”
BackgroundKordestani, H., Xiang, Y., Ye, X., Yun, C., & Shadabfar, M. (2019) · Structural Control and Health Monitoring · doi:10.1002/stc.2491 · Cited by 18 · cited 2×
“Researchers have introduced deep learning models as well as statistical analysis and machine learning as SHM techniques to determine the damaged cable locations [2] or detect stiffness reduction [4].”
BackgroundMethodSecond-order inelastic analysis of steel suspension bridges
Kim, S.-E., & Thai, H. (2011) · Finite Elements in Analysis and Design · doi:10.1016/j.finel.2010.12.007 · Cited by 10 · cited 1×
“In addition, the reliability of PAAP has been evaluated for the cable-stayed bridges [20,21] and suspension bridges [22,23].”
BackgroundSelf-citationEffect of baseline calibration on assessment of long-term performance of cable-stayed bridges
Zhang, J., & Au, F. T. K. (2013) · Engineering Failure Analysis · doi:10.1016/j.engfailanal.2013.01.031 · Cited by 6 · cited 1×
“Thus, the potential damage of cables should be identified early to prevent terrible disasters [48,49].”
BackgroundEFFECT OF DYNAMIC UNLOADING OF CABLES IN COLLAPSE PROGRESSION THROUGH A CABLE STAYED BRIDGE
Das, R., Soumya, A. D. P., Mahesh, M. J., Saini, P., & Anvesh, S. (2016) · Cited by 4
Bumps at Bridge Approaches: Two Case Studies for Bridges at El-Minia Governorate, Egypt
Hassona, F., Hashem, M. D., Abdelmalak, R. I., & Hakeem, B. M. (2017) · Sustainable civil infrastructures · doi:10.1007/978-3-319-61914-9_21 · Cited by 3 · cited 1×
“The quantity of surveyed cables depends on the scale and complexity of the bridge and the monitoring objectives [47,50].”
BackgroundOrder Matters: Sequence to sequence for sets
Vinyals, O., Bengio, S., & Kudlur, M. (2015) · arXiv · doi:10.48550/arxiv.1511.06391 · cited 1×
“The set2set model [45] is permutation invariant for input data using an attention mechanism.”
Method
Research works printed but not linked: 1
- Chen, W.F., and Kim, S.E. (1997). LRFD Steel Design Using Advanced Analysis, CRC Press.
Cited by
Jia, J., & Li, Y. (2023) · Sensors · doi:10.3390/s23218824 · Cited by 182
“…CapsNets to localize and quantify the stiffness reduction damage in beams, showing that the method achieved better results than CNNs. Son et al. [196] proposed that a Graph Neural Network (GNN) can learn graph structure data, which overcame the limitation that a CNN can only learn grid…”
ResultThe application of deep learning in bridge health monitoring: a literature review
Zhang, G.-Q., Wang, B., Li, J., & Xu, Y. L. (2022) · Advances in Bridge Engineering · doi:10.1186/s43251-022-00078-7 · Cited by 102
“Son et al (2021) mapped cable tension to graph vertices and the connection relationship between sensors to its edges, and trained a GNN framework, the message passing neural network, to localize the damaged cables and estimate their area.”
BackgroundChen, Q., Cao, J., & Zhu, S. (2023) · IEEE Internet of Things Journal · doi:10.1109/jiot.2023.3272535 · Cited by 83
“[176] used a message passing neural network (MPNN) to learn the mapping from cables’ tension to the decrease of cables’ section areas.”
MethodRizvi, S. H. M., & Abbas, M. (2023) · Engineering Research Express · doi:10.1088/2631-8695/acefae · Cited by 58
State-of-the-art and annual progress of bridge engineering in 2021
Zhao, R., Zheng, K., Wei, X., Jia, H., Li, X., Zhang, Q., Xu, G., Zhan, Y., Shen, R., Zhang, F., et al. (2022) · Advances in Bridge Engineering · doi:10.1186/s43251-022-00070-1 · Cited by 47
Metadata
- Authors
- Son, H., Pham, V.-T., Jang, Y., & Kim, S.-E.
- Affiliations
- Sejong University (KR)
- Corresponding author
- Jang, Yun
- Venue
- Sensors
- Publisher
- MDPI AG
- Published
- 2021-04-30
- Volume
- 21
- Issue
- 9
- Pages
- 3118
- ISSN
- 1424-8220 (ISSN-L)
- PMC
- PMC8125630
- OpenAlex
- W3159895633
- Cited by
- 29
- References
- 54
- Citations by year
- 2026: 6 · 2025: 7 · 2024: 5 · 2023: 4 · 2022: 7
- FWCI
- 2.07
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Sensors (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- National Research Foundation of Korea (2019R1A4A1021702)
- Topics
- Physical Sciences › Engineering › Civil and Structural Engineering › Structural Health Monitoring Techniques
- Physical Sciences › Engineering › Civil and Structural Engineering › Structural Engineering and Materials Analysis
- Physical Sciences › Engineering › Civil and Structural Engineering › Infrastructure Maintenance and Monitoring
- Keywords
- cable-stayed bridges, damage severity estimation, message passing neural networks, graph neural networks, multi-task learning, structural health monitoring
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.