Find Similar Papers

mosaic similar discovers papers related to a known work — given only its DOI or arXiv ID. No query to write, no keyword to think of: point MOSAIC at a paper you already know and let it map the surrounding literature for you.

mosaic similar 10.48550/arXiv.1706.03762

Similar to: Attention Is All You Need

 #   Title                                       Authors              Year   Source     OA    PDF
 1   BERT: Pre-training of Deep Bidirectional…   Devlin et al.        2019   OpenAlex   yes   ✓
 2   Language Models are Few-Shot Learners       Brown et al.         2020   OpenAlex   no    –
 3   Improving Language Understanding by…        Radford et al.       2018   OpenAlex   yes   ✓
...
10 result(s)

---

How it works

mosaic similar fans out to two independent sources and merges the results:

Source	Method	Requires
OpenAlex	related_works graph — curated semantic neighbours stored per paper	Nothing — always queried
Semantic Scholar	Recommendations API — ML-based similarity model over 214 M papers	ss-key in config

When both sources return the same paper (matched by DOI), the records are merged: the higher citation count wins, and any field missing from one source is filled from the other.

---

Accepted identifier formats

Format	Example
Bare DOI	10.48550/arXiv.1706.03762
doi: prefix	doi:10.1038/s41586-021-03819-2
DOI: prefix	DOI:10.1038/s41586-021-03819-2
arxiv: prefix	arxiv:1706.03762
ARXIV: prefix	ARXIV:2303.08774

---

Options

mosaic similar [OPTIONS] IDENTIFIER

Option	Short	Default	Description
--max	-n	10	Max similar papers to return
--sort		—	citations (most-cited first) or year (newest first)
--oa-only		off	Show only open-access papers
--pdf-only		off	Show only papers with a known PDF URL
--download	-d	off	Download available PDFs
--download-dir		config	Override PDF download directory for this run
--output	-o	—	Save results to file (repeatable); format from extension: .md, .markdown, .csv, .json, .bib

---

Examples

# Basic: 10 similar papers to the Transformer paper
mosaic similar 10.48550/arXiv.1706.03762

# Use arXiv prefix
mosaic similar arxiv:1706.03762

# 20 results sorted by citation count
mosaic similar 10.48550/arXiv.1706.03762 -n 20 --sort citations

# Open-access only, download PDFs
mosaic similar 10.1038/s41586-021-03819-2 --oa-only --download

# Save to BibTeX for Zotero / LaTeX
mosaic similar 10.48550/arXiv.1706.03762 --output related.bib

# Save to multiple formats
mosaic similar arxiv:2303.08774 -n 30 \
  --output related.md --output related.bib --output related.json

---

Enable Semantic Scholar for better coverage

OpenAlex related_works is curated and high-precision but limited in size (typically 20–50 works per paper). Semantic Scholar's recommendation model covers 214 M papers and returns up to 500 candidates, dramatically increasing recall.

Register a free API key at semanticscholar.org and configure it once:

mosaic config --ss-key YOUR_KEY

From that point on, mosaic similar automatically queries both sources and merges the results.

---

Combine with other commands

Use mosaic similar as the first step in a literature-discovery workflow:

# 1. Find related papers, sort by citation count, save to BibTeX
mosaic similar arxiv:1706.03762 -n 30 --sort citations --output seed.bib

# 2. Download open-access PDFs
mosaic similar arxiv:1706.03762 --oa-only --download

# 3. Import everything into a NotebookLM notebook for AI summaries
mosaic notebook create "Transformer neighbours" \
  --from-dir ~/mosaic-papers/