Use the seed paper's similar-articles or recommendation links to generate candidate papers, then keep only those whose abstracts confirm relevance to your topic. A low citation count limits what cited-by and citing-article links can return, so recommendation-based discovery gives you a second route into the literature.
- Open the seed paper's record and locate its recommendation links. In PubMed, use the similar-article links attached to the record. In Google Scholar, open the "Related articles" link for the seed paper. Semantic Scholar's Academic Graph API exposes recommendation endpoints for programmatic retrieval. Expected result: a list of candidate titles that share content or citation patterns with the seed paper.
- Screen the suggested titles against your topic before reading further. Compare each title with your research question and drop anything outside your scope. Expected result: a shortlist of plausible candidates.
- Read the abstract of each shortlisted paper. Add a paper only when its abstract confirms it addresses your topic. Expected result: a set of genuinely relevant papers rather than a long list of loosely connected titles.
- Run a title search on the seed paper itself. Google Scholar supports title searches when the title is placed in quotation marks, which helps you find the paper's indexed record and any related versions. Expected result: additional records and links you can mine for related work.
- Repeat from the strongest new paper. Treat a highly relevant paper from your shortlist as a new seed and open its recommendation or related-article links. Expected result: a wider pool of candidates that stays anchored to your topic.
PubMed also supports field tags, phrase searching, filters, Advanced Search, and citation matching, which you can combine with the recommendation route when you need to narrow results by author, journal, or date. Google Scholar's date sorting and alerts can help you track new work that cites or resembles your seed paper after the initial search.