Skip to content

Research methodology

How Kosmico Deep Research works

Deep Research is a staged research workflow, not a single web search. It discovers papers from registered academic sources, expands and ranks evidence, tracks coverage, and produces artifacts with source records.

The list below says which databases are searched by default, which you can add, which Kosmo adds when the topic calls for it, and which are limited to one field.

Source coverage is a registry, not a slogan

A source count is meaningful only if it says which sources are visible, which are queried automatically, and which are used only for a specific field. This page lists them as the product's own code defines them.

The pipeline separates discovery from synthesis

Candidate discovery, metadata normalization, deduplication, relevance ranking, citation expansion, evidence extraction, coverage assessment, and synthesis are distinct stages. That makes failures and gaps inspectable instead of hiding them inside one generated answer.

A real citation is not automatically supporting evidence

Citation-identity checks ask whether a reference resolves to the claimed paper and has coherent metadata. Claim-to-evidence validation asks a different question: whether the cited passage actually supports the sentence. Kosmico reports these as separate scopes.

Official source

Current product facts

Kosmico v0.7.6 · Verified

Research sources

arXiv · default, on requestSemantic Scholar · default, on requestOpenAlex · default, on requestPubMed · on request, added when relevant, subject-specificCrossRef · on request, added when relevantDBLP · added when relevantEurope PMC · added when relevant, subject-specific

Research pipeline

  1. 01You review the proposal: question, scope and sources
  2. 02Turn the question into research goals and search queries, with a quick web search for context
  3. 03Search the databases in parallel
  4. 04Merge duplicates by DOI, arXiv ID, PubMed ID or near-identical title
  5. 05Rate each candidate's relevance
  6. 06Search again, follow citations or look for gaps, one decision at a time
  7. 07Extract evidence from the full text of the most relevant papers
  8. 08Write the review section by section and build the Knowledge Graph

Stopping conditions

  • It finishes as covered only when it holds the minimum number of papers, its coverage score reaches the target, and every required research goal has supporting papers. The score is computed, not judged by a model: topics, methods, time span, source diversity, citation density and goals.
  • If Kosmo tries to finish earlier, it is sent back to search for the goals still open.
  • Otherwise it stops when the searches or passes allowed by the chosen depth run out, and the result is marked as completed with gaps, not as covered.

Citation identity

Every citation comes from a record fetched from one of the databases, and you can open it. The review may cite only papers found in the run: a section with an unknown reference, or a substantive paragraph without a citation, is rejected and rewritten. An optional pass also resolves each reference by DOI, arXiv ID, PubMed ID or title and flags the ones it cannot confirm.

Claim-to-evidence validation

A claim taken from a paper's full text is kept only if its quoted passage appears on the cited page; claims without a matching quote are dropped. Each section is written from the evidence of the papers it cites, but no model checks that every sentence of the review is supported, so check the sources of anything you are about to publish.

Direct answers

Does Deep Research search the open web?

Not for papers. It runs a quick web search to understand the question, then finds papers only in the academic databases listed on this page, and every citation comes from one of those records.

How does it decide when to stop?

It finishes when it has enough papers, its coverage score reaches the target and every required goal has supporting papers. If the search budget of the chosen depth runs out first, the result is marked as completed with gaps. The benchmark will measure separately whether that decision is well calibrated.

Does a verified DOI prove the related claim?

No. A DOI can establish citation identity. Supporting a claim requires a separate comparison between the claim and evidence in the source.