*
This is my submission for the Hacktoberfest Open-Source AI Challenge — Week 4: Ghost in the Machine.*
What I Built
The internet has changed a lot over the years. Websites that were once popular have disappeared, online communities have become inactive, and technologies that shaped the early internet are slowly being forgotten.
I wanted to build something around this idea.
LOST.INTERNET is a digital archaeology platform that helps people explore forgotten websites, old technologies and different parts of internet culture.
The idea is to make digital history easier and more interesting to discover.
Currently, users can explore a curated archive, filter discoveries by category and search using natural language through AI-powered semantic search.
Demo
Live Website: https://uditsharma45.github.io/lost-internet/
The project is live on GitHub Pages. You can explore the archive and try the search functionality.
Code
GitHub Repository: https://github.com/uditsharma45/lost-internet
The source code is publicly available. Contributions and suggestions are welcome.
How I Built It
I wanted to keep the project lightweight and experiment with open-source AI instead of depending on paid AI APIs.
Here is the technology I used:
- HTML and CSS for the frontend.
- JavaScript for application logic.
- Transformers.js for integrating AI into the browser.
- Xenova/all-MiniLM-L6-v2 for generating sentence embeddings.
- GitHub Pages for deployment.
The semantic search works by converting the user's query and archive descriptions into numerical embeddings and comparing their similarity.
One thing I particularly wanted to explore was running the AI model directly in the browser rather than depending on an external AI API.
The model is downloaded when needed and performs inference in the browser. An internet connection is still needed to load the model initially.
Why Does Open Innovation Matter?
As a second-year Computer Science student, I'm still learning about AI and exploring how these technologies can be used in real-world applications.
Open-source AI makes experimentation much more accessible. It gives developers the opportunity to understand how models work, integrate them into their own applications and build without always depending on expensive commercial APIs.
For this project, it allowed me to experiment with semantic search and browser-based AI inference without requiring a paid AI API key.
I believe this freedom to experiment is one of the most valuable things about open innovation.
The Spirit of It
The inspiration behind LOST.INTERNET is simple: not everything that disappears should be forgotten.
There are countless websites, communities and digital experiences that were once part of people's lives. I find it interesting to think that a part of internet history is still out there, waiting to be explored.
This project is my small attempt to make discovering that history a little easier.
What's Next?
This is an early version, and I have several ideas for improving it:
- Integrating the Internet Archive to explore actual historical website snapshots.
- Expanding the collection of forgotten websites and technologies.
- Improving semantic search and result relevance.
- Allowing users to contribute their own discoveries.
- Adding previews of archived websites.
I'm looking forward to taking this project further.
Built for: Hacktoberfest Open-Source AI Challenge — Ghost in the Machine.
Thanks for reading and checking out my project!
Top comments (0)