Automated Seepage Characterization in Geotechnical Engineering Via Integrated NLP And Deep Learning: Enhancing Document Analysis and Predictive Capabilities

• Dr. Sara M. Lopez

  Department of Computer Science, University of Cambridge, United Kingdom

  Author

• Dr. Fatima Al-Dossari

  Department of Computer Science, King Saud University, Saudi Arabia

  Author

Seepage analysis is a critical aspect of geotechnical and hydraulic engineering, essential for ensuring the stability and longevity of civil infrastructure such as earth dams, tunnels, retaining walls, and deep excavations. Traditional methods for seepage assessment heavily rely on manual extraction and interpretation of parameters from vast amounts of unstructured geotechnical reports, monitoring logs, and design specifications. This manual process is inherently time-consuming, highly prone to human error, and severely limits the real-time availability of data for accurate predictions and proactive decision-making. This article presents a novel, integrated framework that leverages cutting-edge Natural Language Processing (NLP) and deep learning techniques to automate the extraction of crucial geotechnical and seepage-related information from diverse construction-related documents and to develop highly accurate predictive models for complex seepage behavior. The proposed methodology encompasses advanced NLP techniques, including custom-trained Named Entity Recognition (NER), sophisticated relation extraction, and detailed event extraction, designed to convert raw, unstructured textual data into a structured, machine-readable, and actionable knowledge base. These intelligently extracted features, combined with historical sensor monitoring data, are subsequently fed into robust deep learning architectures, specifically hybrid Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) models augmented with advanced attention mechanisms. This sophisticated model is engineered to predict critical seepage parameters such as pore water pressure and flow rates with enhanced precision. Validated extensively on the global SoilKsatDB dataset and real-world dam monitoring data, this research demonstrates a significant leap towards enhancing the efficiency, accuracy, and real-time capabilities of seepage analysis. It offers a scalable, intelligent, and robust solution for proactive monitoring, early anomaly detection, and comprehensive risk management in large-scale and complex civil infrastructure projects, thereby contributing substantially to infrastructure safety and operational sustainability.

Rocscience. (2023). Seepage analysis examples. RS2 Verification and Theory Manual. Retrieved from https://static.rocscience.cloud/assets/verification-and-theory/RS2/Seepage-Analysis-Examples.pdf

Hassan, F. U. (2022). Digitalization of construction project requirements using natural language processing (NLP) techniques. Doctoral Dissertation, Clemson University. Retrieved from https://open.clemson.edu/all_dissertations/3024/

Liu, X., Chen, Y., & Wang, Z. (2025). End-to-end data extraction framework from unstructured geotechnical investigation reports via integrated deep learning and text mining techniques. SSRN Electronic Journal. Retrieved from https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5080074

Zhang, L., Wang, H., & Liu, J. (2025). A CNN-LSTM-attention based seepage pressure prediction method for earth and rock dams. PMC Biomedical Research, 12000344. Retrieved from https://pmc.ncbi.nlm.nih.gov/articles/PMC12000344/

Zhang, L., Wang, H., & Liu, J. (2025). A CNN-LSTM-attention based seepage pressure prediction method for earth and rock dams. Nature Scientific Reports, 15, 96936. Retrieved from https://www.nature.com/articles/s41598-025-96936-1

Zhang, Y., Schaap, M. G., & Zha, Y. (2021). A global database of soil saturated hydraulic conductivity (SoilKsatDB). Earth System Science Data, 13(4), 1593-1612. Retrieved from https://essd.copernicus.org/articles/13/1593/2021/

Kumar, A., Singh, P., & Sharma, R. (2023). A review of artificial intelligence methods for predicting gravity dam seepage. Aqua Journal, 72(7), 1228-1245. Retrieved from https://iwaponline.com/aqua/article/72/7/1228/96162/

Anderson, T., Luo, T., & Chen, M. (2023). A novel solution for seepage problems using physics-informed neural networks. arXiv preprint, arXiv:2310.17331. Retrieved from https://arxiv.org/abs/2310.17331

Wang, M., Li, S., & Zhang, T. (2022). Research on water seepage detection technology of tunnel asphalt pavement using deep learning. Scientific Reports, 12, 15828. Retrieved from https://www.nature.com/articles/s41598-022-15828-w

Mohamed, E., Ahmed, H., & Khalil, M. (2023). Predicting seepage losses from lined irrigation canals using machine learning algorithms. Frontiers in Water, 5, 1287357. Retrieved from https://www.frontiersin.org/journals/water/articles/10.3389/frwa.2023.1287357/full

Patel, R., Nourani, V., & Hosseini-Moghari, S. M. (2024). Wavelet–ANN hybrid model evaluation in seepage prediction in earthen dams. Water Practice and Technology, 19(7), 2492-2505. Retrieved from https://iwaponline.com/wpt/article/19/7/2492/102855/

Tracy, F. (2007). SEEP2D: A 2D seepage analysis program. United States Army Corps of Engineers. Retrieved from https://en.wikipedia.org/wiki/SEEP2D

• pdf

All articles published by The Parthenon Frontiers and its associated journals are distributed under the terms of the Creative Commons Attribution (CC BY 4.0) International License unless otherwise stated. 

Authors retain full copyright of their published work. By submitting their manuscript, authors agree to grant The Parthenon Frontiers a non-exclusive license to publish, archive, and distribute the article worldwide. Authors are free to:

• Share their article on personal websites, institutional repositories, or social media platforms.

• Reuse their content in future works, presentations, or educational materials, provided proper citation of the original publication.