A review of Computing Within Limits - 2015 to 2025

This repository contains code and data for collecting and analysing research published in the conference LIMITS from 2015-2025.

Installation and setup

The following software is required:

• Python
• Grobid. This is only for extracting metadata from papers (step 2).

This project uses uv as project and package manager. Install dependencies with:

uv sync

If you don't have uv, install the dependencies listed in pyproject.toml manually.

Usage

1. Download papers from LIMITS website

uv run python3 src/download_papers.py
uv run python3 src/download_papers.py --debug   # Skip download, only print URLs

IMPORTANT: The script does not work flawlessly (yet). Run in debug mode to improve it.

2. Extract metadata using Grobid

Requires installing and running Grobid.

./src/run_grobid.sh
./src/extract_metadata.sh [FOLDER_WITH_PDFs]

This produces TEI XML files (one per paper) in the xml/ directory.

3. Ingest metadata into database

uv run python3 src/ingest_metadata.py             # Load data into SQLite database
uv run python3 src/ingest_metadata.py --to-csv    # Also export a CSV overview

4. Clean and normalise data

uv run python3 src/clean_papers.py      # Flag excluded papers (slides, off-topic)
uv run python3 src/clean_countries.py   # Resolve author country affiliations

5. Analyse data

uv run python3 src/analyze_papers.py        # Bibliometric analysis and plots
uv run python3 src/analyze_literature_review.py # Manual review analysis and publication-ready figures/tables
uv run python3 src/text_analysis.py         # Full-text term frequency and country analysis
uv run python3 src/create_network_graphs.py # Co-authorship and keyword networks

Output plots and statistics are written to analysis_output/.

PDFs and XML files

This repository contains the PDF and XML files used in the analysis. In order to access that data locally, you need to install Git Large File Storage (Git LFS). Cloning the repository without Git LFS installed will only pull pointer files to the PDF and XML files, and you won't have access to the actual data.

• pdfs: Contains the PDF files downloaded using src/download_papers.py (step 1).
• xml: Contains the XML files generated using src/extract_metadata.sh (step 2).

Progress

• First version of download script.
• Script for extracting metadata using Grobid.
• First version of ingesting metadata into database.
• Script for analyzing data.
•  Gather all PDFs within scope
  • Filter pdfs for false positives. There are some presentations slides there, as well as papers that was not presented at the conference.
  •  Download papers from 2017 and 2018 LIMITS (maunally)
• [-] Rerun metadata extraction with full version of Grobid.
• Add semantic analysis of full body text of research papers