Pub-AI logoPub-AI: AI in Science
Literature Review & Evidence Synthesis⏱ 2–4 hours for initial map; ongoing as the review develops

Mapping a Research Field with Semantic Scholar and ResearchRabbit

A practical workflow for combining Semantic Scholar's search with ResearchRabbit's citation network visualization to rapidly build a comprehensive map of a research field — identifying the landmark papers, key authors, and active frontiers.

AudienceResearchers entering an unfamiliar field or building a comprehensive literature map for a review, thesis, or grant
Tools coveredSemantic Scholar, ResearchRabbit, Zotero
Published July 2026

Why use two tools together?

Semantic Scholar and ResearchRabbit do complementary things:

  • Semantic Scholar is a search tool — you come in with keywords or a question and it returns a ranked list of relevant papers. It’s the entry point.
  • ResearchRabbit is a network exploration tool — you come in with a set of papers and it shows you their citation relationships, surfaces what those papers cite and what cites them, and finds similar work. It’s what you use once you have a seed set.

Using them together lets you start from a keyword search, rapidly expand to landmark papers through citation networks, and identify both foundational work (what the field builds on) and recent developments (who’s citing the landmarks now) — typically in a few hours rather than the days a manual process would take.

Step 1: Build a seed set in Semantic Scholar

Start with 3–5 papers you already know are central to your topic — they might come from a textbook, a supervisor’s recommendation, or a quick Google Scholar search.

If you’re starting with nothing:

  1. Go to semanticscholar.org
  2. Search for your topic using natural language or key terms
  3. Filter by Year (last 5 years for recent state-of-field; expand if you want foundational work) and by Highly Influential to prioritize papers that have had significant citation impact
  4. Open the top results and check: does the abstract describe the core problem you care about? If yes, save to your list.

Target: 5–10 seed papers. More than 10 at this stage makes the next steps unwieldy.

What to look for in Semantic Scholar:

  • Citation count: a highly-cited paper is likely influential, but check the date — a 2024 paper with 50 citations is doing better than a 2020 paper with 50 citations
  • Highly Influential Citations: Semantic Scholar marks these specifically — papers that cite a given work in a way that’s mechanistically important, not just passing references. A paper with many highly influential citations is a field anchor.
  • TLDR summaries: Semantic Scholar provides AI-generated one-sentence summaries — useful for quickly assessing relevance before opening the full abstract

Step 2: Import your seed set into ResearchRabbit

  1. Create a free account at researchrabbitapp.com
  2. Create a new Collection for your topic
  3. Add your seed papers by DOI, title search, or by importing a RIS file from Semantic Scholar or Zotero

Once your seed papers are in, ResearchRabbit builds a visualization showing:

  • The papers themselves as nodes
  • Citation connections between them (which papers cite which)
  • Similar papers it recommends based on the network structure

Step 3: Explore the citation network

This is where the real value is. For each seed paper, ResearchRabbit shows:

Earlier works (what this paper cites)

Click a seed paper → select Earlier Works. This surfaces:

  • The foundational papers that your seed set builds on — these are often the field’s canonical references
  • Methodological papers that introduced key techniques

Add the most relevant ones to your collection. These are your historical anchors.

Later works (what cites this paper)

Click a seed paper → select Later Works. This surfaces:

  • Papers that have built on or applied this work
  • Often includes the most recent developments in the field
  • Can reveal whether the field has moved in directions you weren’t tracking

Add the most relevant recent ones. These represent the active frontier.

Similar papers

ResearchRabbit recommends additional papers based on co-citation patterns (papers that are frequently cited together). This often surfaces closely related work you wouldn’t find via keyword search alone.

Practical tip: process one seed paper at a time. For each one, add the top 3–5 highly relevant earlier and later works before moving on. If you add everything at once, the network becomes overwhelming.

Step 4: Identify the landmark papers

After processing your seed set, look at your collection as a whole. In ResearchRabbit’s network view:

  • Large nodes (many connections) = papers that many in your collection cite = likely field landmarks
  • Papers that appear as Earlier Works of multiple seed papers = foundational references
  • Authors who appear repeatedly across your collection = key researchers in the field

These landmark papers are what your eventual review needs to engage with explicitly. If you haven’t already read them, they’re your next priority.

Step 5: Loop back to Semantic Scholar for gaps

After ResearchRabbit expands your collection, you often identify terms, concepts, or subfields you didn’t know to search for initially. Go back to Semantic Scholar and search these:

  • Specific technique names you encountered in the papers
  • Key author names (Semantic Scholar has author pages showing their full publication list)
  • Related problem framings you didn’t know about

Add any relevant new discoveries to your ResearchRabbit collection. The loop continues until you’re no longer finding papers you haven’t seen before — a sign you’ve reached saturation for your scope.

Step 6: Export to Zotero

When your collection is reasonably complete, export it:

  1. In ResearchRabbit: Export → RIS or BibTeX
  2. Import into Zotero
  3. In Zotero: add tags, organize into collections by subtheme, and attach PDFs for anything you plan to read carefully

From Zotero, your collection is ready for:

  • Systematic screening in Rayyan (if you’re doing a formal review)
  • Deep extraction in Elicit (if you need structured data from full texts)
  • Synthesis in NotebookLM (once you’ve verified your paper set)

What this workflow won’t give you

  • Complete coverage for a systematic review: citation network exploration is not a systematic search. You’ll find the well-connected papers, but you may miss recent preprints, papers in non-mainstream venues, or work in adjacent fields that isn’t heavily cross-cited yet. For a formal systematic review, supplement this with a documented Scopus/PubMed/Web of Science search.
  • Quality assessment: ResearchRabbit finds you papers; it doesn’t tell you which ones are methodologically sound. Read and evaluate the papers you plan to rely on.
  • Access: ResearchRabbit links to papers, but access depends on your institution’s subscriptions. Your library’s Zotero integration or interlibrary loan handles papers behind paywalls.
Purpose Target collection size
Quick field orientation (grant background, intro paragraph) 20–30 papers
Thesis literature chapter 50–150 papers
Scoping review 100–300 papers (before screening)
Systematic review Use Scopus/PubMed for the formal search; ResearchRabbit for initial orientation only

Related content

← Back to Learn