Citation Star Map is a Python-powered interactive scientific literature explorer that transforms research papers into a navigable 3D universe.
Instead of browsing papers through lists and search results, users can explore scientific knowledge as a constellation of interconnected stars, where each star represents a real research paper and every connection represents a verifiable scholarly relationship.
The project focuses on health science, neuroscience, brain connectivity, and network medicine literature while maintaining strict data integrity standards.
Scientific knowledge is a network.
Citation Star Map visualizes that network as an explorable galaxy, allowing researchers, students, and curious minds to discover how ideas connect across disciplines.
Every node is a real paper.
Every edge is traceable.
No fabricated citations.
No synthetic metadata.
- Retrieves papers from scholarly databases
- Uses only real metadata
- Uses only verifiable citation relationships
- Supports DOI and Open Paper links
The graph is built from:
- Direct citations
- Co-citations
If citation coverage is incomplete:
- Keyword co-occurrence relationships are used as a fallback
- Fallback relationships are explicitly labeled
- Louvain clustering
- Automatic research-community discovery
- Color-coded scientific domains
- Orbit navigation
- Zoom controls
- Pan controls
- Auto-rotating camera
- Full-screen immersive visualization
Selecting a paper reveals:
- Title
- Authors
- Abstract
- DOI
- Open access link (when available)
- Data completeness annotation
Users can download:
- Paper metadata
- Citation graph
- References
- Research corpus archive
- Python
- FastAPI
- Pydantic
- NetworkX
- Pandas
- NumPy
- OpenAlex API
- PyAlex
- Crossref API
- Semantic Scholar API (optional)
- NetworkX
- Python-Louvain
- Scikit-learn
Python-first architecture using:
- Reflex
The application minimizes frontend complexity by keeping state management, routing, data processing, and business logic in Python.
- Three.js rendering layer
- WebGL acceleration
- Interactive 3D graph visualization
citation-star-map/
│
├── app/
│ ├── pages/
│ ├── components/
│ ├── state/
│ └── main.py
│
├── builder/
│ ├── fetch_papers.py
│ ├── build_graph.py
│ ├── communities.py
│ ├── layout.py
│ └── export.py
│
├── data/
│ ├── graph.json
│ └── corpus.zip
│
├── requirements.txt
├── README.md
└── LICENSE
✔ Direct citations
✔ Co-citations
✔ Keyword-based fallback edges
✘ Invented citations
✘ Fabricated references
✘ Synthetic papers
✘ Artificial metadata
Every edge must be reproducible from retrieved source data.
Each paper is tagged with one of the following:
| Status | Meaning |
|---|---|
| full_citation_graph | Complete citation information available |
| partial_references | Some references available |
| keyword_fallback | Citation coverage insufficient; keyword relationships used |
The system supports both formats:
authors = "John Doe"and
authors = ["John Doe", "Jane Smith"]through a shared utility:
get_first_author(authors)Application-wide exception handling prevents runtime crashes and blank screens.
All state initialization and hooks execute before conditional rendering.
The graph generation pipeline is deterministic and can be rebuilt from source data.
- Citation analysis
- Literature review
- Knowledge discovery
- Network medicine
- Neuroscience mapping
- Scientific community detection
- Academic exploration
- Multi-domain citation maps
- Temporal citation evolution
- Semantic clustering
- Research recommendation engine
- OpenAlex synchronization
- User-generated collections
- Export to Gephi and GraphML
MIT License
Mapping scientific knowledge as a navigable universe—one citation at a time.