Connected Papers: Understanding a Research Field’s Structure at a Glance

Connected Papers (connectedpapers.com) generates visual graphs of academic papers based on a technique called bibliographic coupling — papers that frequently cite the same sources appear closer together on the graph, even if they don’t directly cite each other. This surfaces relationships that citation chains miss.

How the Graph Works

Start with one paper. Connected Papers builds a graph of 25–50 related papers where: node size represents citation count (larger = more cited), node color represents publication year (darker = older), and edges represent similarity (shared citations). Papers in the same cluster work on similar problems; papers bridging clusters are often methodological connectors.

Reading the Graph

A tight cluster of recent papers around your seed paper means you’ve found an active subfield. A seed paper that bridges two clusters indicates it’s at the intersection of two research traditions — methodologically interesting for novel approaches. Papers at the outer edges of the graph with many connections are often “canonical” background papers that everyone in adjacent fields cites.

Prior Work and Derivative Work

Connected Papers also offers “Prior Work” and “Derivative Work” views. Prior Work shows the foundations your seed paper built on (what it cited most, what those papers cited). Derivative Work shows what built on your seed paper (who cited it most heavily). These two views trace the intellectual lineage of an idea forward and backward in time.

Practical Workflow Integration

Use Connected Papers when you have a good seed paper but don’t know the broader field landscape. Generate the graph, identify the 3–4 main clusters, read one or two central papers from each cluster, and you’ll have a structural map of the field within a day. This works especially well in interdisciplinary research where terminological differences hide conceptual similarities.

Free Tier Limits

Connected Papers allows 5 free graphs per month. For a PhD student, this is often sufficient — you map a new area every few months, not daily. The paid plan ($3–6/month) removes limits and adds PDF integration for papers in your graph.

上一篇 Scite.ai:一篇论文究竟是如何被引用的——支持、反对还是只是提及
下一篇 Connected Papers:一眼看清一个研究领域的结构