Searching the literature

A simple search first

Type plain words into the search box, such as "pea protein gelation hardness", tick the databases to ask, and press Search. No special syntax is needed. Results from several databases are shown as one list; a paper found in more than one database appears once, with the databases that returned it.

Under the results, each database has its own line: how many it found, how many were loaded, or why it failed. A database that is not set up on this platform is named as not set up and is never asked. Nothing is added to the library until you choose papers and press Add.

• Sort by relevance, newest or oldest. Where a database cannot sort that way itself, the loaded results are sorted and the page says so.
• Advanced holds exact phrases, AND, OR and NOT, limits to title, abstract, author, journal or DOI, a year range, protein, functionality, treatment, study type and open access only. For each database the page states which of these it applied and which it could not.
• "Exact query sent to each database" shows what was sent, so a search can be repeated or reported.
• Every search is kept in the history with its date, databases and counts. "Search again" repeats it, and "Save" keeps it as a named search.
• When a count is not given by a database, the page says "at least" rather than showing a total.

Four questions, kept apart

Finding a paper, finding a free copy of it, checking whether automatic download is permitted, and actually getting the file are four separate questions. The table defines each one, and its last column says what you should not conclude from it.

Databases searched for you

PubMed, OpenAlex, Crossref, Semantic Scholar and Europe PMC are public metadata APIs documented for programmatic access. This platform queries them directly, paced to the rate limits each one publishes, and identifies itself by contact email where the provider asks for one.

Results arrive in pages. Each search records how many papers the database reported and how many were actually retrieved, so a partly retrieved search is shown as partial.

Licensed databases, off until your organisation switches them on

Scopus (official API) and Web of Science (official API) are OPTIONAL official APIs, DISABLED BY DEFAULT. Each is enabled only by an operator whose institution holds the entitlement, by setting the connector's switch and supplying the credential SERVER-SIDE. Until then the connector is not constructed and nothing is sent.

These use the official service your institution licenses, with your institution’s own access. The websites of those products are never read automatically.

This platform never scrapes scopus.com, webofscience.com, scholar.google.com, researchgate.net, publisher websites and institutional and library pages. There is no code path that fetches an HTML page from any of them, and none of those hosts appears on any connector's allow-list, so a request to one is structurally impossible.

Databases you search yourself

Some sources cannot be searched for you. Each has its own reason, given below, and each can still be used by searching it yourself and importing the result.

• Google Scholar — Google publishes no search API for Scholar, so there is nothing to connect to. Search it yourself, export the citations or follow a link out, and bring the file to the Imports page. Permitted here: a manual search a person runs in their own browser, a citation export the person downloads and a link-out this platform renders for the person to click.
• ResearchGate — ResearchGate's terms prohibit automated access, so this platform never queries or scrapes it. Open it in your own browser, under your own acceptance of its terms, and bring back a citation file or the DOI. Permitted here: browsing it yourself in your own browser, a link-out this platform renders for the person to click and importing a citation file or a DOI the person brings back.
• MDPI (a publisher, not a database) — MDPI is a PUBLISHER. It publishes journals; it does not index other publishers' work. Searching it answers 'what has MDPI published on this?', which is not the question a systematic search asks, and treating it as a database would bias a review towards one publisher's catalogue. Its articles are indexed by Crossref, OpenAlex, PubMed and Europe PMC, which this platform does query, so the right way to find them is a metadata search that covers every publisher at once. Permitted here: a manual search a person runs in their own browser, a citation export the person downloads from an article page, a link-out this platform renders for the person to click and an open-access PDF the person downloads themselves and uploads here.
• Any other database — Anything with no connector here can still be searched in its own interface and imported as RIS, BibTeX or CSV. Nothing is lost by importing; the search itself is simply recorded as manual rather than executed by this platform. Permitted here: a search a person runs in the database's own interface and importing the exported file here.

Library access stays with you

Access your library holds is access YOU hold. This platform builds an OpenURL to your institution's resolver, a link to its discovery service or the publisher's landing page, and you open it and authenticate yourself. It never authenticates on your behalf and never proxies the request.

This platform never collects, stores, replays or automates a university credential, an MFA response, a browser cookie, a proxy session or any other paywall bypass.

In practice, the platform never asks for your university username or password, a sign-in code or a library proxy session. You open the link, sign in with your own institution, and upload the PDF you obtained.

How a database is added

A source may be added to this registry only when it publishes a documented official API for the access this platform would perform, AND an explicit licensing and security review has been recorded for it. No source is added on the strength of the fact that a request to it happens to succeed.

In short: a database is added only when it offers an official service for this kind of access and the licence has been checked. A source is never added just because a request to it happens to work.

What a search stores

Every search is kept with its exact query, its filters, the database, the number of results reported and retrieved, and the time it ran. A plain-language description of the search is stored too.

This record lets you draw a PRISMA-style flow diagram from stored numbers, and run the same search again later to compare.

Write the search from the protein and property synonyms in the study configuration, so the search and the screening criteria describe the same study.

Advanced options

The optional advanced area adds to the search: extra required groups of terms joined with OR or AND, exclusions written with NOT, and a publication-language filter. Each database receives them in its own syntax, and the preview beside each database shows whether it could apply them.

A database that treats the search as plain words is not sent an exclusion, and several take no language filter. Results from those sources still need screening for what could not be filtered.