ResearchRadar

How ResearchRadar works

From a research question to a traceable reading list. Here is how sources, matching, publication metadata and updates shape your radar.

You choose the question and the sources

A radar combines a name, a search query, selected sources and a freshness window. Source suggestions use topic and keyword matching against the catalogue. You can change that selection: a recommendation does not establish the quality of a source or the completeness of its coverage.

Connectors retrieve available publications

ResearchRadar uses public APIs, RSS or Atom feeds and selected public publication pages. Supported journal connectors search Crossref and OpenAlex by ISSN. Requests have timeouts, rate limits and bounded pagination. An unavailable source can leave gaps; a successful refresh does not imply that every article has been retrieved.

Relevance is a heuristic, not a scientific quality score

The current ranking uses word matches against the query and the position of a result, with percentages generally bounded between 70 and 97. Several connectors also require a minimum number of matching terms. Providers may apply their own search ranking. This is not a trained machine-learning model, a probability of correctness or a measure of a paper’s scientific impact.

Duplicate records and publication versions

Records in the same topic are grouped using normalized DOI, arXiv identifiers, canonical URLs and sufficiently long matching titles. Published-version links can connect a preprint to a journal article. The displayed representative generally favors published work, direct sources and fuller metadata, while respecting saved records. Missing identifiers or changed titles can prevent a match.

The scientific record stays close to the original

Titles and descriptions retain their source language; choosing a UI language does not translate them. The result can show authors, DOI, journal, publication type and a published-version link when available. Metadata can be incomplete or inconsistent across sources. Open the original to review methods, revisions and access conditions.

Freshness, refreshes and saved articles

The chosen time window filters publication dates supplied by sources. Refreshes add or update retrieved records; journal connectors can use incremental synchronization where supported. Dates and indexing delays vary by provider. Saved articles remain in your saved collection even after leaving the radar’s current freshness window.

Briefs, examples and shared radars

Essentials select from retrieved results using relevance and recency; they are a reading shortlist, not a new scientific abstract. Public examples are cached, so they may not change on every visit. Sharing exposes a read-only radar link until its owner disables sharing. A personal copy can then use its own query and source selection.

Use it as a discovery tool

ResearchRadar helps you follow new work across sources. It does not guarantee exhaustive retrieval, full-text access, peer review, reproducibility or the validity of a claim. For systematic reviews, clinical decisions or formal evidence assessments, use the original publications and a documented search and appraisal process.

Explore a radar without an account →