Expert mode search syntax

Advanced search: Expert mode

Introduction to expert mode syntax

This article will cover the specific search syntax that Overton’s advanced search uses in expert mode. Expert mode uses an Embase-style search syntax that helps you build systematic, complex, multi-concept searches.

You can also download our comprehensive user guide, where you’ll find the following information and more.

Prefix

Our expert query builder uses a prefix to tell the system how to interpret your query. This prefix is complex:.

Basic keyword searching

Enter one or more words to find documents containing those terms.

complex:vaccine retrieves documents whose title or description contains vaccine.

Unless you specify a field, we match your terms across the default text fields: the title and translated title, the snippet, and the AI-generated description and theme. We don’t search full text by default.

We don’t use stemming, so searching for nurse won’t automatically match nurses, nursing, or nursed. Use truncation to capture word variants.

Phrase searching

Enclose two or more words in double quotation marks to match them as an exact phrase.

complex:"climate change" retrieves documents containing the phrase climate change (the words adjacent and in that order), rather than the two words appearing separately anywhere in the document.

You can truncate inside a phrase: complex:"climate chan*" matches climate change, climate changes, climate changing, and so on.

When adding a search term, you can choose from the following match types:

  • Keyword: find documents containing these words, however far apart they are
  • Exact phrase: match an exact phrase, with all words in the order given
  • Proximity: match these words within a set number of words of each other, in the order given

You select these using the dropdowns in the builder.

Truncation

An asterisk (*) matches any number of characters at the end of a word.

complex:vaccin* matches vaccine, vaccines, vaccinate, vaccination, vaccinating, and other words beginning with vaccin.

Truncation can be used inside phrases (“climate chan*”) and combined with proximity operators. Wildcards and truncation aren’t supported at the start of a word, for example *nald isn’t valid.

Wildcards

A question mark (?) matches exactly one character. This is useful for spelling variants.

complex:immuni?ation matches both immunisation and immunization. complex:wom?n matches woman and women.

Use ? for single-character variation and * for variable-length endings; the two can appear in the same query.

Boolean operators

Classic Boolean operators are available in the query builder. AND, OR, and NOT work as you’d expect. Search terms can be grouped, see How to use Overton’s expert query builder for more on grouping.

  • complex:vaccine AND immunisation — both terms must be present
  • complex:vaccine OR vaccination — either term may be present
  • complex:vaccine NOT influenza — documents containing vaccine but not influenza
  • complex:(vaccine OR vaccination) AND policy — either vaccine or vaccination, combined with policy. The parentheses ensure the OR is evaluated before the AND.

Combining concepts

Build each concept as a bracketed block, with synonyms joined by OR, then join the concepts together with AND. This is the standard building-block approach used in systematic searching.

complex:("artificial intelligence" OR "machine learning" OR "deep learning") AND (radiology OR "medical imaging")

  • Concept 1: AI-related terms, any of which may appear
  • Concept 2: imaging-related terms, any of which may appear
  • The AND requires at least one term from each concept

You can attach field codes to a whole block, and mix in proximity and truncation freely:

complex:("net zero"~2 OR "carbon budget") AND (transport OR aviation):ti,ab retrieves documents about net zero or carbon budget topics that also mention transport or aviation in the title or snippet.

Proximity operators

Proximity operators find terms that appear near one another, which is far more precise than a plain AND. This is especially useful when searching the full text of policy documents. You can use proximity operators with normal keyword searches and wildcards.

Phrase proximity with ~n

You can apply proximity inside a quoted phrase using a tilde followed by a number, loosening an exact phrase so the words can appear within n words of each other.

complex:"net zero"~2 matches net zero where the two words appear within 2 words of one another.

Phrase proximity with NEAR

NEAR matches terms within a set number of words of each other, in any order.

complex:vaccine NEAR immunisation matches vaccine and immunisation within the default distance (6 words), in either order.

Control the distance with NEAR/n: complex:vaccine NEAR/3 immunisation matches the two terms within 3 words of each other, in any order. The maximum distance is 12 words.

Phrase proximity with NEXT

NEXT works exactly like NEAR, but order matters, the first term must precede the second.

complex:climate NEXT/2 change matches climate followed by change within 2 words.

Combining proximity with truncation and wildcards

Proximity operators work alongside * and ?: complex:carpometacarp* NEAR/3 thumb, complex:vaccin* NEAR immuni?ation.

Cautions

  • Proximity can’t be used with non-text fields, such as publication dates.
  • A Boolean operator can’t sit inside a proximity expression when using AND. For example, ("climate" AND environment) NEAR "net zero" is invalid, since the meaning of AND inside a NEAR isn’t supported. This works when the Boolean operator is OR instead.
  • ADJ4, N3, and similar operators from other platforms aren’t supported. If used, they’re treated as ordinary search words. Use NEAR/n or NEXT/n instead.
  • Proximity operators work between search terms only, you can’t place one next to a field qualifier or a Boolean group.

Field selection and validation

You can select which fields to search in, at both the group level and the row level. If you apply a search field to an entire group, you can only do this for text fields. When applying a field selection to a single row, more fields are available, and you can use the search box to find them.

Some fields require a specific data type, like numbers, for example publication year. If you enter incorrect text, the builder will show an error.

Text field codes

Each field code can be searched differently, depending on its type:

  • Text fields let you search for individual words or phrases; the engine looks for documents containing them.
  • Exact fields require your search term to match the entire value precisely; used for identifiers like external IDs or domain names, where partial matches aren’t helpful.
  • Integer fields hold numeric values and can be searched using exact numbers or ranges.
CodeTypeField
:titextDocument title and translated title
:abtextDocument snippet (sometimes extracted from landing pages)
:aidesctextLLM generated description and theme
:fttextThe full text of the document
:externalidexactA site-specific identifier assigned to an output, similar to a catalogue number or DOI. Only collected for certain sources.
:domainexactFind documents that cite specific domains
:pyintegerPublication year (YYYY). Can be a range, e.g. 2010-2020
:sddateAdded since date. Returns documents since (greater than or equal to) this date (YYYY-MM-DD)
:laexactLanguage of the policy document. Must be ISO693_3, e.g. fre, eng, ger
:cyexactCountry of the policy document source, must match what’s in Index
:sourceidexactPolicy source ID of the policy document source
:sourcesectorexactSector of the policy document source, e.g. Public Sector
:sourcetypeexactType of the policy document source, e.g. Government
:sourcefunctionexactFunction of the policy document source
:sourceauthorexactPolicy organisation/author
:citecountintegerNumber of incoming citations to the policy document, can be a range, e.g. 1-10
:citecountxintegerNumber of incoming citations (including same-source citations), can be a range, e.g. 0-5
:citedrorexactROR that must be cited by policy document results, e.g. https://ror.org/01nrxwf90

Two points worth keeping in mind:

full text is not searched by default, it’s included only when you add :ft. Restricting to title and snippet keeps result sets manageable and searches fast; full-text searches with wildcards and proximity are more demanding and can return very large sets.

To exclude the AI description from a default search, restrict to title and abstract with :ti,ab.

While most of this is covered in our user guide, here are our related Help articles:

Was this article helpful?

Related Articles

Leave a Comment