The problem it solves
Bibliographic searching can lose reproducibility when database selection, each query's syntax, filters or deduplication are not properly documented.
What Vetra does
- Helps select the databases relevant to the question and review type.
- Builds search queries from the full review design.
- Combines thesauri from official sources with keywords, adapting the syntax to each database without losing consistency across strategies.
- Records date, database queried, search string, filters, number of results and strategy version.
- Supports search execution and deduplication of repeated papers across databases.
Human oversight
The AI suggests terms and flags omissions based on the review design. RAISE warns that some assisted search tools are useful as exploratory support, not as a direct substitute for a conventional systematic search.
The researcher can refine the query by talking to the AI and adjust the strategy before running it.
What is recorded
- Selected database and search string with its own syntax.
- Filters, limits, dates and number of retrieved records.
- Duplicate records detected and the deduplication rule applied.
- Changes between versions and justification for the adjustments.
- Exports linked to each run to support PRISMA reporting.
Limits
Vetra builds AI-assisted searches as traceable, reviewable strategies. The goal is to speed up construction and documentation while keeping expert judgment over sensitivity, specificity and coverage.