easy-keywords

Simple utility for extracting some high-salience normalized keywords from a string of English text.

Installation

Install the easy-keywords npm package:

npm install easy-keywords

or

yarn install easy-keywords

Usage

Use the extractKeywords function to extract keywords from a string.

import { extractKeywords } from "easy-keywords"

const keywords = await extractKeywords("the t-shirt is green and nice and soft, it is well liked by both men and women, great success. my dad loved it when he tried it");
// returns an array of strings: [
//   "shirt",
//   "men",
//   "dad",
//   "great",
//   "success",
//   "nice",
//   "green",
// ]

Options:

extractKeywords takes options as a second argument.

• max: Return a maximum of n keywords, like extractKeywords("some phrase", {max: 1}) to get the top scoring keyword from a phrase.

Algorithm

easy-keywords uses TF/IDF to determine term salience, which is a fancy way of saying that terms that occur more frequently in the input text are assumed to be more important. After normalizing each term and identifying which terms have high TF/IDF scores, we filter down to only nouns and verbs to try to get a bit better signal-to-noise.

TF/IDF is the same algorithm that underlies Lucene's and Elasticsearch's idea of search salience and is a good starting point. If you want higher salience, you could put your entire corpus of documents into a search system and use TF/IDF over the whole corpus. This is currently out of scope for easy-keywords but PRs are welcome!