PA-X Peace Agreement Corpus Search

A Streamlit web application for searching and analysing the full PA-X Peace Agreement Database corpus of ~2,257 peace agreements.

Built by PeaceRep, University of Edinburgh.

---

Features

• Keyword & phrase search across full agreement texts, with comma-separated multi-term input
• Boolean search modes — OR (default), AND (terms co-occur in the same sentence), NOT (first term present, remaining absent)
• Exact phrase matching for multi-word expressions
• Word variant matching — approximate stem-based lemmatisation for single-word terms (e.g. internal also matches internally)
• Pre-search filters — region, country/conflict, agreement stage, agreement type, year range, and a three-level topic/issue/sub-issue hierarchy
• Topic-scoped search — optionally restrict search to text segments already tagged for the selected topics
• Sentence-level results — matched snippets with in-text highlighting, per-term counts, and agreement-level summary metrics
• Mention review editor — flag individual sentence-level mentions for exclusion and add notes before export
• CSV export — download all mentions or only reviewed/kept mentions
• Visualisations — charts by year, region, stage, and country, filterable by individual search term
• Per-agreement detail — expandable cards with direct links to PA-X and agreement PDFs

---

Repository structure

corpus_app.py                  # Main Streamlit application
requirements.txt               # Python dependencies
assets/
  logos/                       # PeaceRep and PA-X branding images
pax/
  pax.csv                      # Agreement metadata (region, country, stage, year, links)
  pax_corpus_2257_agreements_v10.csv   # Full text corpus
  only_tagged_pax_topic_codes_segments.csv  # Topic-tagged text segments
archive/                       # Superseded versions (local only, gitignored)

---

Local development

Prerequisites

• Python 3.10+
• pip or uv

Setup

git clone https://github.com/peacerep/corpus_app.git
cd corpus_app

python -m venv .venv
# Windows
.venv\Scripts\activate
# macOS / Linux
source .venv/bin/activate

pip install -r requirements.txt

Run

streamlit run corpus_app.py

The app will open at http://localhost:8501.

---

Data files

File	Description
pax/pax.csv	Agreement-level metadata: title, year, region, country, stage, agreement type, PA-X and PDF hyperlinks
pax/pax_corpus_2257_agreements_v10.csv	Full agreement texts keyed by AgtId
only_tagged_pax_topic_codes_segments.csv	Text segments tagged with PA-X topic codes, used for topic filtering and topic-scoped search

Data source: Bell, C., & Badanjak, S. (2019). Introducing PA-X: A new peace agreement database and dataset. Journal of Peace Research, 56(3), 452–466.

---

Deployment on Streamlit Cloud

Initial deploy

1. Push the repository to GitHub (e.g. github.com/peacerep/corpus_app).
2. Go to share.streamlit.io and sign in with your GitHub account.
3. Click New app and configure:
   • Repository: peacerep/corpus_app
   • Branch: main
   • Main file path: corpus_app.py
4. Click Deploy.

Updating a live deployment

Streamlit Cloud automatically redeploys when changes are pushed to the configured branch:

git add .
git commit -m "your message"
git push origin main

The app will rebuild and restart within a minute or two.

Forcing a clean rebuild

If the app is showing a stale version or dependency errors after a push:

1. Go to the app in the Streamlit Cloud dashboard.
2. Click the ⋮ menu (top-right of the app tile) → Settings.
3. Under General, confirm the Main file path is set to corpus_app.py.
4. Return to the dashboard, click ⋮ → Delete cache and reboot.

A plain Reboot restarts the process but reuses the cached package install. Delete cache and reboot forces a full fresh install from requirements.txt.

Secrets

This app does not currently use any API keys or secrets. If secrets are needed in future, add them via Settings → Secrets in the Streamlit Cloud dashboard and access them with st.secrets.

---

Dependencies

Package	Purpose
streamlit	Web application framework
pandas	Data loading and manipulation
plotly	Interactive charts
nltk	Snowball stemmer for word-variant matching
spacy	NLP pipeline (en_core_web_sm model)

See requirements.txt for pinned versions.

---

Licence

Data: PA-X Peace Agreement Database — University of Edinburgh.
Code: Niamh Henry, PeaceRep, University of Edinburgh.