arxiv-sanity-lite

Why we included this project

Keeping up with the daily flood of new machine learning papers on arXiv is genuinely hard, and this project attacks that problem directly. It periodically pulls new papers from the arXiv API, then lets you tag the ones you care about and get similar-paper recommendations computed with SVMs over TF-IDF vectors of the abstracts. The bundled Flask web UI handles searching and ranking the results, and an optional cron job emails you daily suggestions for each tag. It is a deliberately small, lightweight re-write of the older arxiv-sanity, which makes it practical to run your own instance on a cheap VPS and index tens of thousands of papers. Anyone who wants a personal, self-hosted research radar without relying on a third-party service will find this a straightforward, hackable foundation.