Large Language Model for Geotechnical Engineering Applications Using Retrieval Augmented Generation
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
The application of Natural Language Processing (NLP) in the field of Civil and geotechnical engineering presents unique challenges with the complexity of the calculations and the necessity for engineering judgment. Large Language Models (LLM) have been successfully presented to be one of the most effective solutions in many specific fields such as arts and literature, finance, and even software development. However, the use of LLM in the field of Civil Engineering and specifically in Geotechnical Engineering has not been widely discussed often compared to the other fields. This paper aims to cover the details with the framework and breakdown of how LLMs can be used to retrieve information from domain-specific data set specifically related to geotechnical engineering. The paper covers topics ranging from settlement analysis, ground improvements, to scripting features with the Rocscience platform. Data sources include research papers, technical documents from geotechnical engineers, and documentation for online help from the relevant geotechnical software. The paper is structured into three main sections: an introduction to LLM models with a focus on pre-trained models and model selection; a discussion of the data retrieval process using domain-specific database using Retrieval Augmented Generation (RAG); and an evaluation of the model’s performance. Performance benchmarks are established with vector embedding to trace context retrieval, and validation is performed using methods such as Recall-Oriented Understudy for Gisting Evaluation (ROGUE), Bidirectional Encoder Representations From Transformer (BERT), and Bilingual Evaluation Understudy (BLEU) score to compare the result with prompt-completion pairs. The study demonstrates that an LLM with sufficient geotechnical domain-specific data source produces superior responses compared to LLM trained on a generic public data set. The paper concludes with a discussion on the comparative results and the future potential of the LLM in geotechnical engineering.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it