This project lists ideas for how to get updated and hear of new papers. Feel free to use any of the tools and to share new tools that you created\found\use.
Start here — what do you want to do?
| I want to… | Go to |
|---|---|
| See what’s hot right now across the field | What’s trending |
| Follow specific people / topics and get a feed tuned to me | Build a personal feed |
| Make sure I don’t miss anything in my area | Cover a whole field |
| Save and organize what I’ve read | Keep track of what you’ve read |
| Find papers related to one I already have | Find related papers |
Rule of thumb: use trending/feed tools to discover, search/related tools when you have a specific paper or question, and a reference manager so you never lose a paper twice.
What’s trending
Best for a quick pulse on the field — what people are reading and talking about this week.
- alphaXiv — trending arXiv papers with public discussion threads.
- Twitter/X — the fastest signal for new work. Follow field-specific lists rather than individuals to cut noise (example lists in NLP and ML).
- Bluesky — a smaller, more focused academic community (strong in ML/NLP). Less volume than Twitter, but higher signal-to-noise, and you can subscribe to curated feeds:
- Paper Skygest — a personalized feed of just posts that link to papers, ranked by your follow network.
- LinkedIn — some researchers now post papers here too. Currently the noisiest and least focused of the three for research, but worth a glance if your field is active on it.
- labml — surfaces the papers Twitter is buzzing about, ranked.
- trendingPapers — a site dedicated to what’s trending.
- Papers with Code — Trending — trending papers, weighted toward those with released code.
Tips
- Trending ≠ important — it favors flashy and recent. Use it to *notice*, not to decide what matters. - Following a couple of good curated lists beats following 200 accounts.Build a personal feed
Best for staying current on your own topics — set it up once, then let recommendations come to you.
Follow specific researchers
- Twitter/X — follow authors to catch work as they announce it.
- Semantic Scholar — follow authors and get recommendations in your library feed.
- ResearchGate — author-following (less popular in computer science).
- Google Scholar — get email alerts for a key author’s new and related work (how-to).
Get a feed tuned to your taste (learns from what you like over time)
- Scholar Inbox — a daily digest that adapts to your ratings. Good default for a “morning coffee” feed.
- Research Rabbit — recommendations built from collections you assemble.
- Semantic Scholar — recommendations from the papers in your library.
- Zeta Alpha — neural search + recommendations aimed at AI/ML.
- PaperParrot — personalized new-paper alerts.
Tips
- Taste-based feeds are cold at first — rate a handful of papers up front and they improve quickly. - Pick **one** feed tool and stick with it; running several just duplicates your inbox.Cover a whole field
Best when you can’t afford to miss anything in your area — comprehensive, at the cost of some noise or delay.
- Main conference proceedings — the highest-quality filter, though it arrives months after the work appears.
- arXiv subscriptions — get every new submission in chosen categories by email. Comprehensive but high-volume.
Tips
- arXiv-by-category is a firehose — pair it with a trending or taste feed so you actually read the important ones.Keep track of what you’ve read
Best for never losing a paper, and for citing it later without re-hunting.
Reference managers (store, tag, cite; browser + Word/LaTeX integration)
- Zotero — free and open-source; the common recommendation for students.
- Mendeley — reference manager with a built-in PDF reader and social features.
- EndNote — established, often provided through university licenses.
Lighter-weight options
- Semantic Scholar Folders — save papers and get notified when similar new ones appear.
- Shared Google Doc or GitHub list — public, editable, and easy to share, though not a convenient long-term library.
- General note apps
- Obsidian — markdown notes linked to papers, extensible with plugins (and AI tooling such as Claude Code).
- Notion — flexible database-style tracking of a reading list.
Tips
- Whatever you pick, install its **browser connector** — one-click saving is what makes you actually keep it up. - A reference manager pays off most at writing time: it generates your bibliography for you.Find related papers
Have a paper (or a question) and want its neighbors? Pick by how you want to explore.
Follow the citation graph (visual — “what’s near this paper”)
- ConnectedPapers — one seed paper → a similarity graph of its neighborhood. Fast way to map an area.
- litmaps — like ConnectedPapers, but you build and track a living map over time.
- Research Rabbit — graph exploration plus ongoing recommendations.
Ask a research question (AI search over the literature)
- Asta / AI2 paper finder — natural-language search tuned for scientific papers.
- Google AI-based search — experimental natural-language search over Scholar.
- Elicit — ask a question, get papers with extracted findings in a table. Best for evidence-gathering.
- Sourcely — AI paper search with summaries.
Search classically (keyword / author / who-cited-whom)
- Google Scholar — broadest coverage; use “cited by” for forward citations.
- Semantic Scholar — cleaner metadata, citation context, and an influential-citation signal.
- Citation Gecko — start from seed papers and snowball through their citations.
Low-tech but effective
- Read the related-work / background / intro of a good recent paper — often the fastest curated map of an area.
Tips & caveats
- ConnectedPapers: use an incognito window if you go past the free-use limit; paid plans for heavy use. - Citation-graph tools are great for *breadth*; a single well-written related-work section is great for *judgement about what matters*. - ([Elicit was formerly at elicit.org](https://elicit.org/) — the current site is [elicit.com](https://elicit.com/).)Hosted here: git repo, in this branch — feel free to contribute.