Tools Search Strategy Builder
Search Strategy Builder

Build the search string for your systematic review and get the query ready for each database

A five-step guided wizard separates Blocks A, B and NOT from your topic and writes the query in the syntax of fourteen databases, from Scopus to LILACS.

Version
2026.08
Last updated
Available since

Free credits · No credit card · Instant access What changed recently

What this solves

Writing a search string by hand means keeping fourteen different syntaxes straight: [tiab] in PubMed, TITLE-ABS-KEY in Scopus, TS= in Web of Science. The Search Strategy Builder asks for your production type, research areas, topic, what is in and out of scope, and the time frame, then returns the conceptual blocks and one query per database, ready to copy.

How it works

  1. Choose your starting path

    The landing page offers two paths. The guided chat is for a topic that is not settled yet: six short questions turn into three research paths with objective, scope and suggested databases, and you pick one to refine. The direct wizard is for a defined topic and known databases.

  2. Answer the five steps of the wizard

    Production type and title, research areas (27 in the catalog, which pre-select the usual databases for each one), topic and institutional context, what you want and what you do not want in the results, and finally the time frame plus the databases.

  3. Copy the query for each database

    The result shows Block A (central concept), Block B (methodological contribution) and Block NOT (exclusions), one query per selected database with its syntax note, screening criteria and an operational checklist. Copy database by database or download the whole strategy as a text file.

Method

What happens to your file, step by step, and where the result stops being a decision made by a model.

How it works inside

Your wizard answers (production type, title, research areas from the Scopus ASJC list of 27, theme, scope, exclusions, period, databases) are validated by deterministic code and assembled into a single request. Descriptors for the selected databases only (field operators such as [ti]/[tiab], TITLE-ABS-KEY, TI=/TS=, the period operator and syntax notes) come from a hard-coded catalog of 14 databases and go into the same prompt. GPT-4o (temperature 0.4), set to reply only with a JSON object, returns the strategy, which is then rendered into TXT.

Where AI is used, and where it is not

The model writes the substance: Block A synonyms, Block B method markers, exclusions and the query string for each database. Deterministic code validates input (areas, theme length, period normalization to the current year), supplies the catalog syntax, applies a syntax fix (quoting multi-word phrases inside Scopus TITLE and TITLE-ABS-KEY) and renders the document. Database names, URLs and access type come from the catalog, never from the model.

What it accepts

  • Expert flow: production type, title, at least one of the 27 Scopus ASJC research areas, a central topic of at least 10 characters, optional context, target venue and included/excluded scope, start year (defaults to 2020) and optional end year (defaults to the current year).
  • Databases: any subset of the 14 in the catalog (PubMed, Scopus, Web of Science, IEEE Xplore, ACM, SciELO, ERIC, dblp, Embase, PsycINFO, CINAHL, Cochrane, JSTOR, LILACS). If you select none, a fixed area-to-database mapping picks them for you.
  • Free-text databases outside the catalog are accepted and forwarded to the model with a flag instructing it to warn that the syntax is approximate.
  • Discovery flow: subject and research objective, both at least 10 characters. Answers such as 'não sei' or 'no idea' are rejected by a normalized blacklist before any call.

What you get back

  • Structured JSON: objective, included and excluded scope, period, methodological notes, Block A, Block B, Block NOT, one query per database with notes, at most 2 suggested extra databases with justification, screening criteria (INCLUIR / EXCLUIR / REVISAR) and an operational checklist.
  • A rendered TXT with Markdown-style headings: the query in a code block per database, plus the URL and access type (gratuito, capes, assinatura, parcial) taken from the local catalog.
  • Each strategy is saved to your history as its own record and can be downloaded again later as a .txt file.
  • Discovery flow: up to 3 candidate topics, each with a working title, theme, rationale, suggested research areas and suggested databases.

What this tool does not do

  • It does not run the search. No query is ever sent to PubMed, Scopus or any other database, so no result count validates the strategy before you paste it yourself.
  • It does not verify descriptors. MeSH, Emtree, DeCS and thesaurus terms are written by the model and are checked against no controlled vocabulary.
  • Only the Scopus query receives a deterministic syntax fix. Queries for the other 13 databases are delivered exactly as the model wrote them, including the period operator.
  • For a database you typed in yourself, the query comes from the model's recollection of that platform and is flagged as approximate syntax.
  • The focus is the first draft of the strategy. Formalizing a systematic review stays with you: screening the results and documenting reproducibility. The screening criteria it returns come as text for you to apply.
Plan
Pro
Cost
15 credits

What you get

  • Fourteen databases, each with its own syntax

    PubMed/MEDLINE, Scopus, Web of Science, IEEE Xplore, ACM Digital Library, SciELO, ERIC, dblp, Embase, PsycINFO, CINAHL, Cochrane Library, JSTOR and LILACS. Each query uses that database's field operators, such as [tiab], TITLE-ABS-KEY and TS=, plus its year filter.

  • Blocks A, B and NOT kept separate

    The central concept, the methodological contribution and the exclusions come as separate term lists, so you can adjust one block without rewriting the whole string.

  • Two starting points

    If the topic is still open, six guided questions return three deliberately different research paths with objective, scope and suggested databases. If it is already defined, go straight to the wizard.

  • Screening criteria and checklist

    Beyond the queries, the file brings methodological notes, a screening criteria table with decision and rationale, and an operational checklist. It downloads as a text file and stays in your history for later.

Changelog

Every line below is a change that actually shipped, dated by the day it went out.

  1. Latest
    • The guided chat now opens with the subject already filled from the active project's research question. It still asks, nothing is sent automatically, and one keystroke clears it.
    • Leaving the end year blank no longer produces an open range that Web of Science rejects: an absent end year now closes at the current year.
    • The active project stays selected when you move between tools and result pages.
Show 5 earlier updates
    • Tools opened inside a research project now use that project's research question when assembling the strategy.
    • Generated strategies are filed under the project that produced them.
    • The per-database query boxes and the full strategy preview now follow the light theme instead of staying dark.
    • Tool pages share the same title treatment, hero and primary button, with icons and titles corrected for the light theme.
    • Scopus queries: phrases with two or more words inside TITLE and TITLE-ABS-KEY are wrapped in double quotes before the result is shown.
    • When an answer in the guided chat is too short, the chat points at the exact question to redo, in Portuguese and in English.
    • Redesign in three screens: path choice on the landing page, a six-question guided chat, and a five-step wizard with copy per database.
    • The landing page lists the most recent strategies for download again.
    • Full interface in English and Portuguese, with validation messages that say which answer needs to be redone.
    • Access to the tool is checked against the subscription plan on every screen.
    • First version of the Search Strategy Builder: catalog of academic databases, conceptual blocks and one query per database.
    • Generated strategies are saved in a history you can download again.
    • Generation ships behind a feature flag and stays turned off until it is enabled in the environment.

Questions and answers

Does the tool run the search in the databases?

No. It writes the strings and you paste each one into the matching database, logged in through your institution or CAPES when the base requires it. It also does not replace the review protocol: the generated terms need your validation before you register the search.

Which databases are covered, and what if mine is not on the list?

The catalog has fourteen databases with their field operators already mapped. There is also a free field where you can type other names, for example BASE or Lens. The tool still builds a query for them, but it marks a warning that the syntax is approximate and needs to be checked on the platform itself.

Do the queries come out in Portuguese?

No. Search strings are always generated in English, even when the interface is in Portuguese. The explanations, syntax notes and screening criteria that the AI writes follow your interface language; the short fixed usage notes that come from the database catalog stay in English. For SciELO and LILACS, which index Portuguese and Spanish, review the terms and complete them in those languages before running the search.

Which plan do I need and what does each generation cost?

The tool is available from the Pro plan on. Each complete strategy consumes 15 credits and the topic brainstorm in the guided path consumes 5. Strategies already generated stay in the history and can be downloaded again without spending credits.

How to cite this tool

Used it in your research? Here is the reference, already filled in with the version you are looking at and today's access date.

ABNT (NBR 6023)

LESSA, P. W. B. Search Strategy Builder. Versão 2026.08. [S. l.]: Xplore Dados, 2026. Disponível em: https://xploredados.com/en/tool/search-strategy-builder. Acesso em: 21 ago. 2026.

APA 7

Lessa, P. W. B. (2026). Search Strategy Builder (Version 2026.08) [Computer software]. Xplore Dados. https://xploredados.com/en/tool/search-strategy-builder

BibTeX

@software{xploredados_search_strategy_builder_2026,
  author  = {Lessa, Patrick Wendell Barbosa},
  title   = {Search Strategy Builder},
  organization = {Xplore Dados},
  version = {2026.08},
  year    = {2026},
  url     = {https://xploredados.com/en/tool/search-strategy-builder},
  urldate = {2026-08-21}
}

Start with the free credits

Create an account and test the tools before deciding on a plan.

Create free account