Unearthing the Ghosts of the Early Internet with Local AI 🕯️
Description and use
Web Séance is a digital archaeology tool that uses a local Gemma 3:4b model to analyze hyperlinks in archived early‑internet pages, preserving evidence and separating facts from hypotheses.
Problem
Investigate forgotten corners of the early internet using archived webpages while keeping evidence and interpretation separate.
Outcome
Web Séance enables systematic exploration of early internet culture by automating evidence extraction and providing AI-assisted interpretation while maintaining traceability and evidential integrity.
Workflow
- Retrieve historical captures via Internet Archive CDX API
- Parse archived HTML with BeautifulSoup to extract outgoing hyperlinks and assign Evidence IDs
- Normalize and classify links
- Visualize hyperlink trails in a D3.js network graph
- Query a locally running Gemma 3:4b model through Ollama to interpret evidence
Evidence
Web Séance is an AI-assisted digital archaeology tool for exploring forgotten internet culture through historical webpages preserved by the Internet Archive.
Retrieve historical webpage captures using the Internet Archive's CDX API.
BeautifulSoup is used to parse archived HTML and extract outgoing hyperlinks.
Gemma 3:4b, running locally through Ollama, helps interpret the evidence.
The application visualizes observed hyperlink relationships in a D3.js network graph.
- Record date: 2026-11-02T08:29:08.017Z