Monitor New AI Papers on arXiv
Create a scheduled arXiv paper monitor with category and topic searches, revision-aware identifiers, and low-noise first-seen alerts.
Create separate saved Tasks in the arXiv Papers Scraper for category streams such as cs.LG and focused phrases such as retrieval augmented generation. Sort by submission date for discovery and by last-updated date for revision monitoring.
Store a versionless paper key, the returned arXiv ID, and both timestamps. Alert when a paper key is first seen; route updated versions to a separate revision queue. This prevents a revision from looking like an entirely new paper.
Keep alert volume bounded and review title/abstract samples. AI terminology moves quickly, so version the query taxonomy instead of silently changing it. Every alert should link to the arXiv abstract page for source context.
Frequently asked questions
Should the monitor use topic text or categories?
Use both in separate Tasks: categories give coverage while topic expressions provide precision.
How are revisions handled?
Compare the versionless paper ID and updated timestamp, while retaining the returned versioned ID.