Starlight ML

A lightweight JavaScript machine learning utility library for text processing and simple ML tasks. Designed for use in Starlight scripts, Node.js, or browser environments with ES modules (.mjs).

Features

• Tokenization – Split text into words or custom tokens.
• Stopword removal – Filter common words from text.
• Bag-of-Words – Count frequency of words.
• Mapping & utilities – Create dictionaries and perform simple text-to-vector transformations.
• ESM-ready – Can be imported as an ES module.

Installation

Node.js

npm install starlight-ml

Browser (via jsDelivr)

<script type="module">
  import * as ml from 'https://cdn.jsdelivr.net/npm/[email protected]/starlight-ml.mjs';

  const text = "Hello world! This is Starlight ML.";
  const tokens = ml.tokenize(text);
  console.log(tokens);
</script>

Usage

Import in Node.js

import * as ml from "starlight-ml";

const text = "Machine learning is fun!";
const tokens = ml.tokenize(text);

console.log(tokens); // ["Machine", "learning", "is", "fun"]
console.log(ml.removeStopwords(tokens));
console.log(ml.bagOfWords(tokens));

Functions

• tokenize(text: string): string[] – Split text into words.
• removeStopwords(tokens: string[]): string[] – Remove common stopwords.
• bagOfWords(tokens: string[]): Record<string, number> – Count occurrences of words.
• mapTokens(tokens: string[], func: Function): any[] – Map over tokens with a custom function.

License

MIT © Dominex Macedon