Where the papers come from
Which databases Ponder searches, and where full texts come from.
Every paper Ponder finds comes from a public academic literature database and can be traced back to its original record. This article explains where those records come from, so you can judge the coverage of a search.
The main databases: OpenAlex, PubMed and Semantic Scholar
Ponder searches these three standard academic databases, which together hold more than 200 million records across every discipline.
- OpenAlex: an open catalogue of literature across all disciplines, and Ponder's primary database.
- PubMed: the literature database for medicine and the life sciences. The scope at the bottom left of the input box defaults to All; switch a medical or life-sciences question to PubMed and the search runs inside it alone, giving more focused results.
- Semantic Scholar: a multidisciplinary database, used for supplementary searching.
Supplementary searching: more languages and more academic sites
During Analyze and Draft, Ponder searches on its own wherever the material is thin. Beyond the three databases above, it also runs semantic searches on preprint platforms such as arXiv and bioRxiv, and on Korean academic databases such as DBpia, RISS and KISS. This supplementary pass fills in literature in languages other than English, notably Chinese and Korean, so that research can reach your material too.
Where full texts come from
Once Ponder has a paper's record, it looks for an open-access version through several channels: the publisher's open version, PubMed Central, Europe PMC, arXiv and others.
For open-access papers Ponder can read the full text. For papers behind a subscription or paywall it reads the abstract and bibliographic data, and those papers carry an "Abstract only" badge when they are handed to Analyze or Draft. For papers you have access to through your school or institution, upload the PDF into the project and Ponder can read the full text — see Import your sources.
Where bibliographic data comes from
Every paper is identified by its DOI or by its record in a database. Authors, year, journal and the rest of the bibliographic data are taken from that record, and the in-text citations and reference list are generated from them when you write. That is also what makes citations re-typesettable by style and checkable in one click — see Citations and tracing claims.
What to read next
- To see what Ponder does to keep conclusions traceable: read Responsible AI and limitations
- To start from a question and find literature: read Find the most relevant, usable papers
- To see how your data is stored and protected: read Your data and privacy