Had this challenge for a job interview.
Algorithm does a single pass breaking on ngram length and using built in js functions which are pretty optimized already.
While I could've used a Tree based approach I opted a stack based solution using indexes.
During my tests this was much more performant but more benchmarking should be done. I assume it has to do with CPU cache being used much more effectively.
Finally it was specifically asked for a single file private gist with short installation instructions. This is all at index.js and the reason this project has a minimal package.json and no .gitignore
Project is dependency free (except for Jest required to run the tests)
Make sure you already have
node >12
npm
git
Two options:
(Recommended)
$ git clone https://github.com/UnGaucho/ngram-challenge
$ cd ngram-challenge
$ npm install
$ jest
Or (not recommended)
Use the provided instructions at index.js
Typing "node ngram.js" followed by gram length as a number followed by your text
Example:
node ngram.js 3 Whatever you'd like ngramed
N-grams are a contiguous sequence of n words from a string of text, and have many applications from full-text search indexes to machine learning.
You'll generate a set of every permutation of contiguous n-grams from a string of text, from 1-gram to n-grams where n is the number of words in the string
- Program in the language for the position for which you are applying
- Strip punctuation
- Include tests
- Write a single code file that includes both the application and tests As comments in the code file, include installation instructions & instructions for how to run your test and run your code Applicaiton accepts a parameter to generate n-grams from the parameter
"Hello there admiral Kenobi."[
"Hello",
"Hello there",
"Hello there admiral",
"Hello there admiral Kenobi",
"there",
"there admiral",
"there admiral Kenobi",
"admiral",
"admiral Kenobi",
"Kenobi"
]