This repository contains a Python-based classification pipeline for large-scale, reproducible classification of academic literature in structural engineering.
The code was developed to support systematic mapping and review studies of AI applications in finite-element–based structural analysis, design, and automation. It processes BibTeX files and associated PDFs, enriches each entry with structured classification fields, and exports summary statistics for downstream analysis.
- Automated processing of BibTeX + PDF corpora
- Token-aware PDF ingestion for large documents
- LLM-based multi-label classification
- Dynamic parallelism with token-per-minute (TPM) monitoring
- In-place enrichment of BibTeX entries
- Per-folder and global CSV summaries
- Scales to thousands of papers
This repository supports the methodology presented in:
Vaktskjold, V. E., Toppe, L. O., Luczkowski, M., Rønnquist, A., Morin, D.
Systematic Mapping of Artificial Intelligence Applications in Finite-Element-Based Structural Engineering
Buildings, 2026.
https://www.mdpi.com/2075-5309/16/3/644
If you use this code in academic work, please cite the paper.