Connected Papers
Visual graph of papers related to a seed paper, built from similarity and citation patterns — useful for discovering seminal work and derivative papers you didn't know to search for.
What it does
Connected Papers takes a single seed paper (by title, DOI, or URL) and generates a visual graph of related work. Papers appear as nodes; proximity in the graph indicates relatedness, not direct citation. The algorithm combines citation patterns with co-citation similarity — papers that are cited together often appear close to each other even if they don’t directly cite one another.
The graph is split into two panels:
- Prior work — foundational papers that influenced the seed, grouped by publication era
- Derivative works — more recent papers building on the seed paper’s ideas
This approach is different from a standard citation search (which only shows direct links) and different from keyword search (which finds papers using the same words, not necessarily the same ideas).
Best for
Getting an overview of a research landscape around a specific paper — particularly useful when you’ve found one excellent paper and want to discover the intellectual neighborhood it sits in. Also good for identifying the canonical “must-cite” papers in a subfield you’re new to, and for spotting recent work that builds on a seminal paper you’ve already read.
Pricing
Freemium. The free tier allows 5 graphs per month — sufficient for occasional use. The paid tier ($6/month) is unlimited and required for active literature work. No institutional license option; individual billing only.
Strengths
- Surfaces related work you wouldn’t find through keyword search — particularly valuable for interdisciplinary work where the same concept appears under different terminology across fields
- The prior/derivative split gives you temporal context — you can immediately see whether a paper is foundational (heavily prior-cited) or a recent extension
- Visual layout makes cluster structure in a research area immediately legible
- Powered by Semantic Scholar’s data, so coverage of computer science, biology, and physics is strong
- Each node links directly to the paper’s Semantic Scholar page for abstract and open-access full text
Limitations
- 5 graphs/month on the free tier is a real constraint for systematic searching
- Coverage depends entirely on Semantic Scholar’s index — humanities, social science conference papers, and some regional journals are underrepresented
- The similarity algorithm occasionally surfaces unexpected results in sparse fields with few papers; the graph is less reliable when the seed paper is very recent (few citations yet) or in a very narrow niche
- Not suitable as a substitute for a full systematic search — it samples rather than exhaustively covers a topic
How it compares
| vs. | Key difference |
|---|---|
| ResearchRabbit | ResearchRabbit uses pure citation networks and allows you to build a growing library over time; Connected Papers uses similarity-weighted graphs and is better for a one-shot “neighborhood” view of a single paper |
| Litmaps | Litmaps emphasizes chronological evolution of a field; Connected Papers emphasizes semantic proximity across the whole literature |
| Semantic Scholar | Semantic Scholar’s “References” and “Cited By” tabs show direct links; Connected Papers shows the broader similarity neighborhood |
| Citation chaining in Scopus/WoS | Manual forward/backward citation searches are exhaustive; Connected Papers is faster for an overview but less complete |
Practical notes
Start with the most canonical paper in your area — the one everyone cites — rather than the most recent. A well-cited seed generates a denser, more useful graph. For a recent paper with few citations, try the paper’s most influential reference as the seed instead.