Research Discovery Engine (IntelliDE)

The Research Discovery Engine is a collection of research prototypes and supporting analyses for exploring scientific knowledge as structured, connected information. The repository currently includes:

• DE/: a React, TypeScript, Vite, and Tailwind frontend for browsing the Discovery Engine knowledge graph and experimenting with concept design and LLM-assisted workflows.
• path/: a Python network-analysis pipeline over the linked Markdown knowledge base, with checked-in outputs under path/results/.
• resnei/: a small Flask-based ResNei prototype.
• website_explore_the_unknown/: a Next.js website prototype.
• Docs/: the numbered Discovery Engine system-design documentation.
• ResNei SDD/: the ResNei software design document and diagrams.

The repository is experimental and contains multiple related prototypes rather than one installable package. Read the component README before running a component.

Quick start: Discovery Engine frontend

git clone https://github.com/ActiveInferenceInstitute/Research-Discovery-Engine.git
cd Research-Discovery-Engine/DE
npm install
npm run dev

Open the local URL printed by Vite (normally http://localhost:5173). The frontend reads the knowledge base from DE/KG/ at runtime.

To run the automated launcher instead:

python3 main.py

For the frontend's detailed setup, LLM configuration, scripts, and troubleshooting, see DE/docs/README.md.

Analysis pipeline

The Python analysis pipeline is run from path/:

cd path
python -m venv .venv
source .venv/bin/activate  # Windows: .venv\Scripts\activate
pip install -r requirements.txt
python research_paths.py

It reads the Markdown knowledge-base files in path/ and writes staged artifacts beneath path/results/. See path/README.md for the current output layout.

Design documentation

The system-design series is indexed by Docs/0. DE-Paper-Outline.md. It describes the proposed Conceptual Nexus Model, template system, synthesis pipeline, FAIR goals, limitations, assumptions, and future direction. These documents describe the design and research direction; they should not be read as a claim that every proposed capability is implemented in the current prototypes.

Configuration and credentials

The root .env.example documents configuration used by the Python tooling. Keep credentials in an untracked .env file and never commit API keys. The DE frontend uses Vite-prefixed variables such as VITE_API_PROVIDER, VITE_OPENAI_API_KEY, and VITE_OPENROUTER_API_KEY; see DE/src/llm/config/LLMConfig.ts for the current names.

Contributing

See CONTRIBUTING.md for the scope of contributions and the local validation expectations.

License and citation

Repository materials are licensed under CC BY 4.0. The repository has a Zenodo DOI: 10.5281/zenodo.15084931.

For the institute, see Active Inference Institute. The public project prototype is also presented at explore-the-unknown.vercel.app.