Skip to content

Delve Indexer syntax reference

Updated by Matthew

Query syntax

Reference for the boolean query language used in the indexer.

Searching for words and phrases

What you write

What it does

tesla

Matches the word “tesla”.

"electric vehicle"

Matches the exact phrase, in order.

electr*

Matches any word starting with “electr” (electric, electronics, etc.).

Use double quotes around anything with spaces, punctuation, or special characters.

The * wildcard normally only works at the end of a word. Certain fields (currently only domain) may also allow it in the middle or at the start (e.g. *thing , som*ing ).

Combining searches

Operators must be written in uppercase. Lowercase and/or/not are treated as ordinary words.

Operator

Example

Meaning

AND

tesla AND layoffs

Both must appear.

OR

tesla OR ford

Either may appear.

NOT

tesla NOT lawsuit

Excludes results matching the second term.

( ... )

(tesla OR ford) AND layoffs

Groups parts of a query together.

If you combine different operators in the same query, you must use parentheses to make the grouping clear. For example, tesla AND ford OR gm is not allowed — write (tesla AND ford) OR gm or tesla AND (ford OR gm).

Proximity search: NEAR

NEAR finds two terms that appear close to each other in the text.

What you write

What it does

tesla NEAR layoffs

The two words appear within 5 words of each other (the default).

tesla NEAR/10 layoffs

Within 10 words of each other.

(tesla OR "elon musk") NEAR/8 layoff*

Either left-hand option appears within 8 words of the right-hand wildcard.

The terms on either side of NEAR must be plain words, phrases, wildcards, or a simple OR list — not full sub-queries with AND or NOT.

Searching a specific field

Put a field name and a colon before a term to restrict the search to that field.

What you write

What it does

title:tesla

Matches “tesla” in the title only.

site:"Wall Street Journal"

Matches articles from that publication.

content:(tesla NEAR layoffs)

Applies the proximity search to the article body only.

wordCount:>100

Matches articles with more than 100 words.

< and > prefixes can only be used with numerical fields (wordCount, lnSourceRank).

Available fields:

Field

What it matches

title

The article headline.

titleCS

The article headline (case-sensitive)

content

The article body.

contentCS

The article body (case-sensitive)

site

The publication name (e.g. “Reuters”).

domain

The publication’s web domain (e.g. reuters.com). Supports wildcards anywhere in the value.

e.g., to search anything on a Yahoo domain:

(domain:yahoo.com OR domain:*.yahoo.com)

NOTE: You should generally use Delve’s built-in publication list filters instead, for most/all cases you’d want to reach for this field.

author

The article author.

country

The country of publication (two-letter country code).

NOTE: You should generally use Delve’s built-in country filters instead, for most/all cases you’d want to reach for this field.

language

The article language (two-letter language code).

NOTE: You should generally use Delve’s built-in language filters instead, for most/all cases you’d want to reach for this field.

mediaType

Can be ONLINE, BLOG, PRINT, BROADCAST, or FORUM (or empty, for other categories)

wordCount

The word count (e.g. wordCount:>100)

lnSourceRank

An internal rank of the publication value from one of our data providers; from 1 (highest value) to 5 (lowest value). We do NOT control this value, so please use with caution, and understand that we may deprecate it in the future in favor of our own ranking.

e.g, to just see sources ranked 3 or higher: lnSourceRank:<4

url

The article URL.

NOTE: This often points to an internal URL created by Delve or our data partners; you probably want to use domain instead.

(You may see references to other fields like source or sourcePublicationId - you can ignore those, as they’re for internal filter use and not meant to be used directly).

If no field is specified, the search runs against the title, body, and URL.

Comments

Anything after !! on a line is ignored, up to the end of the line. Useful
for annotating long multi-line queries.

!! exclude wire copy NOT site:examplewire AND ( !! ticker and synonyms TSLA OR Tesla* OR title:"elon musk" )

Worked example

NOT site:examplewire AND ( TSLA OR Tesla* OR title:"elon musk" OR content:(layoffs NEAR/3 (factory OR plant)) ) AND NOT ( "press release" OR (sponsored NEAR content) )

This query:

  • Excludes anything from the publication examplewire.

  • Requires one of: the ticker TSLA, any word starting with Tesla, the
    headline phrase "elon musk", or layoffs appearing within 3 words of
    factory or plant in the article body.

  • Also excludes anything containing the phrase "press release" or the word
    sponsored near the word content.