PaperLink

Turn any collection of research papers into an interactive 3D galaxy visualization.

Live Demo (PaperGoat Galaxy) · Live Demo (Apple ML)

---

Features

• 3D Interactive Globe — Papers as glowing nodes on a sphere, edges as curved lineage arcs
• Click & Explore — Click any node for details, Shift+Click for multi-select
• Filter & Search — Filter by category, year, score; full-text search
• Dark / Light Mode — Automatic system preference detection
• Bilingual (EN/KR) — Built-in i18n with one-click toggle
• Responsive — Desktop, tablet, and mobile optimized
• Single HTML — No server, no build tools required for basic usage
• GitHub Pages Ready — Deploy in 2 minutes

---

Quick Start

Option A: Fork & Deploy (2 min)

1. Fork this repository — Click the "Fork" button above
2. Enable GitHub Pages — Go to Settings > Pages > Source: main / root
3. Done! — Your galaxy is live at https://<your-username>.github.io/paperlink/

The template ships with 20 sample papers. Edit papers.json to add your own.

Option B: Local Development

# Clone
git clone https://github.com/jsonpassion/paperlink.git
cd paperlink

# Copy sample data
cp papers.sample.json papers.json

# Open in browser (any local server works)
python -m http.server 8000
# → http://localhost:8000

---

Customizing Your Data

Step 1: Edit papers.json

Each paper needs these fields:

{
  "id": "unique-id",
  "rank": 1,
  "alias": "Short Name",
  "title": "Full Paper Title",
  "url": "https://arxiv.org/abs/...",
  "year": 2024,
  "score": 95,
  "categories": ["LLM", "Scaling"],
  "cluster": "llm",
  "reason": "Why this paper matters (shown in detail panel)",
  "has_review": false,
  "one_liner": "",
  "key_discoveries": [],
  "review_url": "",
  "reason_en": "English version of reason"
}

Step 2: Define Edges (Paper Lineage)

Edges show relationships between papers:

{
  "edges": [
    {"source": "paper-A", "target": "paper-B", "type": "lineage"}
  ]
}

Step 3: Customize Clusters

Clusters define the color-coded categories. Customize them in the clusters section:

{
  "clusters": {
    "llm":        {"label": "LLM",          "color": "#3b82f6", "icon": "🧠"},
    "vision":     {"label": "Vision",       "color": "#10b981", "icon": "👁️"},
    "generation": {"label": "Generation",   "color": "#f59e0b", "icon": "🎨"}
  }
}

Step 4: Build for Production (Optional)

To inline data into a single HTML file for deployment:

python build.py
# → dist/index.html (single file, no external data fetch)

Options:

python build.py --input my_custom_data.json
python build.py --output build/index.html

---

Scoring Rubric

Papers are sized proportionally to their score (80–100 range). The default scoring criteria:

Weight	Criterion	Description
30%	Paradigm Shift	Did it open a new research direction?
25%	Adoption & Industry	Real-world production usage
20%	Citation & Ecosystem	Downstream research impact
15%	Technical Novelty	Originality and depth
10%	Timeliness	Current relevance and trajectory

---

Project Structure

paperlink/
├── index.html              ← Visualization engine (Three.js)
├── papers.json             ← Your paper data (gitignored in production)
├── papers.sample.json      ← Sample data with 20 mock papers
├── build.py                ← Inline papers.json into HTML for deployment
├── dist/
│   └── index.html          ← Production build (after running build.py)
├── LICENSE                 ← MIT
└── README.md

---

Keyboard Shortcuts

Key	Action
ESC	Reset selection & filters
Shift + Click	Multi-select nodes
Scroll	Zoom in/out
Drag	Rotate globe

---

Live Examples

These sites are built with the same engine:

• PaperGoat Galaxy — Top 151 AI papers, curated & ranked
• PaperGoat Apple — Top 100 Apple ML research papers

---

Contributing

PRs are welcome — bug fixes, new features, UX improvements. Please open an issue first for large changes.

---

License

MIT License. See LICENSE for details.

Built by Jason Lee as part of the PaperGoat project.

If you find this useful, a star would be appreciated!