🔬

Connected Papers

Visual literature maps for better research discovery

Free | Freemium | Paid | Enterprise ⭐⭐⭐⭐☆ 4.4/5 🔬 Research & Learning 🕒 Updated
Visit Connected Papers ↗ Official website
Quick Verdict

Connected Papers is a visual research-map tool that builds graph-based literature maps from a seed paper or topic, ideal for researchers and PhD students who need to discover relevant papers and citation neighborhoods quickly. It excels at relationship visualization rather than full-text search, and its pricing includes a usable free tier with paid plans for larger graph exports and private projects.

Connected Papers is a research & learning tool that builds visual, graph-based literature maps from a seed paper, author, or DOI to reveal connected works and citation neighborhoods. Its primary capability is generating undirected co-citation and similarity graphs that surface papers you might miss with keyword search. The key differentiator is interactive, time-aware graph layouts and lineage views that reveal prior and derivative works. Connected Papers serves researchers, graduate students, and R&D teams who need literature discovery and mapping rather than full-text access. Pricing is accessible with a free tier and paid plans for larger graphs and private projects.

About Connected Papers

Connected Papers launched as a specialized literature mapping web app focused on visualization and citation relationships rather than replacing bibliographic databases. Founded by Tzu-Ming (Tom) Kuo and his team, the product positions itself as a discovery layer sitting on top of bibliographic metadata and citation graphs to help users visually explore a research field. Its core value proposition is making the structure of a literature corpus visible: you drop in a seed paper or DOI and Connected Papers computes a similarity map, so you can spot clusters, seminal works, and peripheral but relevant papers without crafting complex search queries.

The app provides several concrete features for exploring a topic. Graph generation produces an undirected similarity graph (up to the plan limit) showing nodes for papers and weighted edges for relatedness; nodes include title, authors, year, and a link to PDFs when available. The “Lineage” view extracts chronological predecessors and successors to trace the development of an idea. The “Document View” surfaces abstracts, references, and citation counts inline. Users can export graph data (GraphML/CSV) and PNGs for presentations; higher-tier accounts raise node limits and unlock private projects and higher-resolution exports. The site integrates DOI search, arXiv links, and CrossRef/semantic metadata sources to populate node details.

Connected Papers has a freemium model. The free tier lets users create a limited number of graphs with modest node counts (the public “small” graph limit), lets you view lineage and basic exports, and allows public saves. Paid plans start with an individual paid tier that increases node limits, enables private graphs and high-res PNG/GraphML exports, and provides more daily graph generations. Team/Institutional or custom enterprise pricing unlocks larger node counts, SSO, and priority support. Exact monthly prices and node limits change periodically; the site lists current plan names and prices on its pricing page and offers month-to-month billing or annual discounts.

Researchers, PhD students, and R&D teams commonly use Connected Papers for literature reviews, grant preparation, and onboarding into a new subfield. For example, a PhD candidate uses it to map 50–150 papers to find methodological clusters and gaps; a translational medicine researcher uses it to identify seminal clinical-to-preclinical lineage papers. The tool is complementary to reference managers and search engines—think of Connected Papers as the visual discovery layer; users aiming for full-text ingestion or citation analytics at scale may pair it with tools like Semantic Scholar or Scopus for deeper metrics and coverage.

What makes Connected Papers different

Three capabilities that set Connected Papers apart from its nearest competitors.

  • Undirected similarity graph and Lineage view focus on co-citation/relatedness rather than keyword matches, surfacing non-obvious connections.
  • GraphML and CSV export for integration into network analysis workflows separates it from tools that only provide visual screenshots.
  • Private graphs and higher node-count mapping on paid plans enable reproducible mapping for teams and reproducible figures in publications.

Is Connected Papers right for you?

✅ Best for
  • PhD students who need comprehensive literature maps for thesis chapters
  • Academic researchers who require visual citation neighborhoods for grant proposals
  • R&D scientists who need to identify methodological lineages and seminal works
  • Librarians and research managers who curate reading lists and field overviews
❌ Skip it if
  • Skip if you need full-text search and ingestion across paywalled journals.
  • Skip if you require large-scale citation analytics with proprietary citation counts and altmetrics.

✅ Pros

  • Visualizes citation/similarity neighborhoods that reveal clusters missed by keyword search
  • Lineage view helps trace idea development through predecessors and successors
  • GraphML/CSV export enables downstream network analysis in Gephi or Python

❌ Cons

  • Coverage depends on available metadata; some paywalled or non-indexed papers may be missing
  • Not a replacement for full bibliographic databases—limited analytics (no built-in large-scale citation metrics)

Connected Papers Pricing Plans

Current tiers and what you get at each price point. Verified against the vendor's pricing page.

Plan Price What you get Best for
Free Free Limited public graphs, low node count, basic PNG exports only Casual users exploring single papers
Individual $9.99/month Higher node limit, private graphs, GraphML and high-res PNG exports Independent researchers and students
Pro $29.99/month Larger graph limits, more daily generations, priority exports Active researchers and consultants
Team Custom SSO, shared workspace, higher node and export limits Labs and institutional teams

Best Use Cases

  • PhD student using it to map 50–150 papers and identify three research clusters
  • Research scientist using it to find seminal predecessors and five key methods papers
  • Librarian using it to assemble a 30-paper reading list for onboarding new faculty

Integrations

arXiv CrossRef DOI/Publisher links

How to Use Connected Papers

  1. 1
    Enter a seed paper or DOI
    Paste a DOI, title, or arXiv ID into the home search box and press Enter; success looks like the seed paper appearing as the central node in the generated graph.
  2. 2
    Generate the similarity graph
    Click the Generate Graph button (or Create graph) to compute the undirected similarity map; wait for the progress indicator — a network of nodes and edges will render on completion.
  3. 3
    Inspect Lineage and Document View
    Click any node to open Document View and then select Lineage to see predecessors and successors; success is seeing abstracts, references, and year-based lineage links.
  4. 4
    Export or save the graph
    Use the Export button to download GraphML/CSV or high-res PNG (export available per plan); or click Save and choose Public/Private to store the project in your account.

Ready-to-Use Prompts for Connected Papers

Copy these into Connected Papers as-is. Each targets a different high-value workflow.

Generate 30-Paper Overview Map
30-paper overview from a single seed
Role: You are Connected Papers assistant creating a concise literature overview. Constraints: Use the current seed paper/DOI selected in Connected Papers; return a co-citation/similarity graph of ~30 most relevant papers; prioritize diversity across methods, datasets, and years. Output format: 1) Short summary (2–3 sentences) of the seed paper's neighborhood; 2) A numbered list of 30 papers with title, year, DOI, one-line reason for inclusion; 3) Suggested 3-entry reading order (starter → method → advanced). Example: "1. Smith et al. (2018) DOI... — foundational method for X."
Expected output: A short neighborhood summary, numbered list of 30 papers with metadata and one-line reasons, plus a three-step reading order.
Pro tip: If the seed is very new, expand to include one highly-cited predecessor per cluster to avoid overly narrow maps.
Locate Seminal Methods Papers
Find seminal predecessors and core methods
Role: Act as a literature scout focused on methods lineage. Constraints: Use the current seed paper/DOI; return up to 12 seminal predecessor papers (older, highly-cited, method-defining) and up to 8 derivative method papers (applications/extensions). Prioritize methods that directly enable the seed's approach. Output format: Two labeled lists — "Seminal predecessors" and "Key derivative methods" — each entry: title, authors, year, DOI, 15–20 word justification. Example entry: "Klein et al. (2005) DOI... — introduced algorithm A used by subsequent model architectures."
Expected output: Two labeled lists (seminal predecessors and derivative method papers) with metadata and 15–20 word justifications for each entry.
Pro tip: Ask Connected Papers to highlight papers that repeatedly co-occur across different citation neighborhoods to surface truly foundational methods.
Build 3-Cluster Literature Map
Map 50–150 papers and identify clusters
Role: You are a research analyst producing a mid-size literature map. Constraints: Use the current seed paper/DOI; generate a graph sized between 50 and 150 nodes; identify exactly 3 coherent clusters; assign a concise label for each cluster; within each cluster list the top 5 representative papers (title, DOI, year) and one-sentence rationale. Also produce a recommended reading order per cluster (3 steps) that respects prerequisite knowledge. Output format: JSON object with keys: "graph_size", "clusters" (array of 3 objects with name, top_papers, reading_order), "notes" (2–3 caveats).
Expected output: A JSON object describing graph size, three clusters with top 5 papers and one-sentence rationales, plus a 3-step reading order per cluster and brief caveats.
Pro tip: If clusters are uneven, rebalance by merging the smallest cluster into the nearest neighbor and note the merge reason in "notes".
Curate 30-Paper Onboarding List
Curate 30-paper onboarding reading list
Role: Act as an academic librarian assembling an onboarding reading list for a new faculty member. Constraints: Starting from the current seed paper/DOI, produce exactly 30 papers balanced across four subtopics (list the subtopics you choose), ensure at least 40% are from the last 5 years, include at least five historical/seminal works, and limit to one paper per research group if alternatives exist. Output format: CSV-compatible table with columns: rank, title, authors, year, DOI, subtopic, citation_count (if available), 20–30 word recommendation note. Example row: "1, Smith et al., 2019, DOI..., Subtopic A, 320, Good survey covering X."
Expected output: A CSV-compatible table of 30 papers with columns including subtopic and short recommendation notes, balanced per constraints.
Pro tip: To ensure pedagogical flow, order each subtopic's papers from survey/introductory to technical/deep-dive rather than purely by citation count.
Produce Grant-Focused Literature Lineage
Create lineage view and gap analysis for grant
Role: You are a senior PI and research strategist using Connected Papers to draft the "prior work and gaps" section of a grant. Multi-step constraints: 1) Using the current seed paper/DOI, produce a lineage view limited to 100 nodes highlighting prior foundational works and immediate derivatives; 2) Identify 3 concrete knowledge gaps or unresolved limitations (each supported by 2 citations from the graph); 3) Propose 3 specific, testable research questions addressing those gaps; 4) Recommend 5 potential collaborators (name, affiliation, 1-line rationale linked to a paper). Output format: JSON with keys: "lineage_summary" (3–4 sentences), "gaps" (array with citations), "research_questions", "collaborators". Example collaborator entry: "Dr. X, Univ Y — expert in Z (see DOI...)".
Expected output: A JSON object containing a 3–4 sentence lineage summary, three gaps each with supporting citations, three testable research questions, and five collaborator recommendations with rationales.
Pro tip: Prioritize collaborators whose recent papers appear in the seed's immediate derivative cluster to increase likelihood of complementary expertise.
Map Method Variants and Adaptations
Identify transferable techniques across domains
Role: You are an interdisciplinary research scientist mapping method variants across domains. Multi-step: 1) From the current seed paper/DOI, identify up to 8 distinct methods or algorithmic variants present in the neighborhood; 2) For each method, list 3 representative papers (title, DOI, domain); 3) For each method provide a 3-step actionable adaptation recipe to transfer it to a different target domain; 4) Suggest one practical evaluation metric for each adaptation. Output format: JSON array of method objects: {"method_name","representative_papers":[...],"adaptation_recipe":[step1,step2,step3],"evaluation_metric"}. Few-shot examples: {"method_name":"Domain Adaptation A","representative_papers":["Lee 2018 DOI..."],"adaptation_recipe":["pretrain on X","fine-tune with Y"],"evaluation_metric":"F1 on held-out domain"}.
Expected output: A JSON array of up to 8 method objects each with representative papers, a 3-step adaptation recipe, and one evaluation metric.
Pro tip: When suggesting adaptations, always include a minimal data requirement (e.g., ~100 labeled examples) to make the recipe actionable for practitioners.

Connected Papers vs Alternatives

Bottom line

Choose Connected Papers over ResearchRabbit if you prioritize exportable GraphML graphs and lineage views for academic figure production.

Head-to-head comparisons between Connected Papers and top alternatives:

Compare
Connected Papers vs Rill Data
Read comparison →

Frequently Asked Questions

How much does Connected Papers cost?+
Monthly plans start around $9.99; Team/Enterprise pricing is custom. The free tier remains available with limited graph sizes. Paid individual plans increase node limits, enable private graphs, and unlock GraphML and high-resolution exports. Teams pay custom rates for SSO, shared workspaces, and larger node/export quotas—check the site pricing page for current exact monthly and annual rates.
Is there a free version of Connected Papers?+
Yes — there is a free tier with limits. Free accounts can generate a small number of public graphs with lower node counts and basic PNG exports. Free users can explore Lineage and Document View but will hit node and daily-generation caps; private graphs and high-resolution/GraphML exports require a paid plan.
How does Connected Papers compare to ResearchRabbit?+
Connected Papers emphasizes visual similarity graphs and lineage exports; ResearchRabbit focuses on recommendation feeds and citation tracking. Use Connected Papers if you need GraphML exports and lineage mapping for figures, and pick ResearchRabbit if you want continuous recommendation feeds and collaborative reading lists.
What is Connected Papers best used for?+
Connected Papers is best for visual discovery and mapping citation neighborhoods. It helps users surface clusters, seminal works, and methodological lineages when starting a literature review. The tool is especially useful for mapping 30–150 papers into clusters and exporting the graph for presentations or further network analysis.
How do I get started with Connected Papers?+
Start by entering a DOI, paper title, or arXiv ID in the search box and generating a graph. Use the rendered network to click nodes, open Document View for abstracts, then toggle Lineage to trace predecessors; save or export the graph when satisfied.

More Research & Learning Tools

Browse all Research & Learning tools →
🔬
Perplexity AI
Research & Learning AI with fast, cited answers
Updated Mar 26, 2026
🔬
Elicit
Automated literature workflows for research & learning
Updated Apr 21, 2026
🔬
SciSpace
AI research assistant for faster literature understanding
Updated Apr 22, 2026