Systematic paper recommendation and discovery using multiple methods
Finding the right papers to read is a research skill in itself. Beyond keyword searches, modern researchers have access to a rich ecosystem of recommendation tools that use citation networks, semantic similarity, co-authorship patterns, and collaborative filtering to surface relevant papers you might otherwise miss.
This skill provides a systematic approach to paper discovery that goes beyond passive reading. It covers algorithmic recommendation services, citation-based discovery techniques, social and community-driven methods, and strategies for building and maintaining a well-curated reading pipeline. The goal is to minimize the chance that you miss an important paper while avoiding information overload.
Whether you are entering a new field and need foundational papers, tracking the frontier of a mature research area, or looking for interdisciplinary connections, this guide provides concrete methods for each scenario.
OpenAlex provides concept-based and citation-based discovery for 250M+ works across all disciplines:
# Find works related to a specific paper via its concepts and citations
curl "https://api.openalex.org/works?filter=cites:W2741809807&sort=cited_by_count:desc&per_page=10"
Use OpenAlex's concept graph to find related work by browsing papers tagged with the same research concepts, or trace citation networks to find derivative and foundational papers.
Connected Papers (connectedpapers.com) builds a visual graph of papers related to a seed paper. It uses co-citation and bibliographic coupling analysis rather than direct citation links, which means it can surface related work even when two papers do not cite each other directly. Use this when:
Google Scholar's "Related articles" feature and the personalized recommendation emails (if you maintain a Google Scholar profile) use a combination of citation analysis and content similarity. To maximize their usefulness:
Research Rabbit (researchrabbitapp.com) lets you build collections of papers and then visualizes networks of related work, similar work, and suggested papers. It integrates with Zotero for importing existing libraries. Key features:
When algorithmic tools are insufficient, manual citation-based techniques remain powerful:
Start with a foundational paper. Find all papers that cite it (using Google Scholar, OpenAlex, or Web of Science). Screen these citing papers by title and abstract to find relevant descendants. Repeat for the most important descendants.
Read the reference list of a key paper. Identify and retrieve the most important cited works. This traces the intellectual lineage of ideas and helps you find the seminal papers in a subfield.
Two papers that are frequently cited together in other papers are likely related, even if they do not cite each other. Tools like VOSviewer and CiteSpace can visualize co-citation clusters from a set of papers, revealing the intellectual structure of a field.
Two papers that share many references are likely addressing related questions. This is the inverse of co-citation and is more useful for discovering recent papers that have not yet accumulated citations.
A sustainable paper discovery practice requires more than one-off searches. Build a pipeline that continuously surfaces new relevant work:
Avoid the trap of an ever-growing, never-read paper queue:
Finding papers outside your primary field is particularly challenging because you may not know the right terminology. Strategies include: