//  include the Keyword Extractor
const keyword_extractor = require("keyword-extractor");

//  Opening sentence to NY Times Article at
/*
http://www.nytimes.com/2013/09/10/world/middleeast/
surprise-russian-proposal-catches-obama-between-putin-and-house-republicans.html
*/
const sentence =
"President Obama woke up Monday facing a Congressional defeat that many in both parties believed could hobble his presidency."

//  Extract the keywords
const extraction_result =
keyword_extractor.extract(sentence,{
    language:"english",
    remove_digits: true,
    return_changed_case:true,
    remove_duplicates: false

});

/*
  extraction result is:

  [
        "president",
        "obama",
        "woke",
        "monday",
        "facing",
        "congressional",
        "defeat",
        "parties",
        "believed",
        "hobble",
        "presidency"
    ]
*/

The module can also be imported with ES module syntax, e.g. in Next.js apps or other ESM-based projects:

import keyword_extractor from "keyword-extractor";
// or, using named imports:
// import { extract, getStopwords, supported_language_codes } from "keyword-extractor";

const extraction_result = keyword_extractor.extract(sentence, {
    language: "english",
    remove_digits: true,
    return_changed_case: true,
    remove_duplicates: false
});

TypeScript

This package ships its own type declarations, so no @types package is needed. Usage is the same, but with import/typed options:

import keyword_extractor from "keyword-extractor";

const sentence = "President Obama woke up Monday facing a Congressional defeat.";

const extraction_result: string[] = keyword_extractor.extract(sentence, {
    language: "english",
    remove_digits: true,
    return_changed_case: true,
    remove_duplicates: false
});

Options Parameters

The second argument of the extract method is an Object of configuration/processing settings for the extraction.

Parameter Name	Description	Permitted Values
language	The stopwords list to use. ISO 639-1 codes and verbose names	ar, cs, da, de, en, es, fa, fr, gl, it, ko, nl, pl, pt, ro, ru, sv, tr, vi, arabic, czech, danish, dutch, english, french, galician,german, italian, korean, persian, polish, portuguese, romanian, russian,spanish, swedish, turkish, vietnam
remove_digits	Removes all digits from the results if set to true (can handle Arabic and Perisan digits too)	true or false
return_changed_case	The case of the extracted keywords. Setting the value to true will return the results all lower-cased, if false the results will be in the original case.	true or false
return_chained_words	Instead of returning each word separately, join the words that were originally together. Setting the value to true will join the words, if false the results will be splitted on each array element.	true or false
remove_duplicates	Removes the duplicate keywords	true , false (defaults to false )
return_max_ngrams	Returns keywords that are ngrams with size 0-integer	integer , false (defaults to false )
stopwords_regex	Removes any word matching one of the given regular expressions, in addition to the language's stopwords list. Useful for filtering out patterns that aren't in a fixed word list, such as units (`10lbs`, `10Kg`), times (`6pm`), or ranges (`1-100`)	Array of `RegExp` (defaults to [] )

Each pattern is tested against both the lowercased and original-case forms of a word, so patterns work whether or not they include the i flag.

Removing words with `stopwords_regex`

const extraction_result = keyword_extractor.extract("He weighed 10lbs at birth and grew to 10Kg, waking at 6pm daily.", {
    language: "english",
    stopwords_regex: [/^\d+[a-z]+$/i]
});

//  extraction_result is ["weighed", "birth", "grew", "waking", "daily"]

Releases

This package is released to npm automatically using semantic-release. Every merge to main is analyzed and, if it contains a releasable change, published under the appropriate version bump.

Versioning is driven entirely by commit messages (and, by extension, pull request titles — PRs are typically squash-merged, so the PR title becomes the commit message on main). Pull request titles must follow the Conventional Commits specification and are checked automatically by CI:

<type>[optional scope]: <description>

Common types and the version bump they trigger:

Type	Description	Release
`fix`, `perf`, `revert`	A bug fix, performance fix, or revert of a previous commit	Patch
`feat`	A new feature	Minor
`refactor`, `docs`, `style`, `test`, `build`, `ci`, `chore`	No user-facing change	None
Any type with a `BREAKING CHANGE:` footer or a `!` after the type/scope (e.g. `feat!:`)	Breaking API change	Major

Release notes for each version are published on the GitHub Releases page — there's no CHANGELOG.md committed to the repo, since main is a protected branch and the release automation can't push commits to it.

Credits

The initial stopwords lists are taken from the following sources:

Keywords

text keyword search extract