mispell

README

This application transforms text into something that sounds more like something a "bimbo" might write. This is done by inserting some words with little meaning like "like", "um" and "basically" into reasonable places in sentences using natural language processing. Words are also misspelled using a curated list of misspellings and some rules (like replacing tt with t in words). The curated list of misspellings is also used to replace words with synonyms that sound more bimbo-ish, or mean something slightly different (ex: then -> than). In addition, numbers like sixty-three, or 1238, or 15003 are able to be scrambled and rounded.

The frequency of replacement and misspelling is decided by a "bimbofactor" value, ranging from 0 to 1. This value can be even higher, which will increase the frequency more, but 1 is balanced to be what sounds maximally bimbo-like while maintaining some level of readability.

Example:

Thiz aplication transformz text into something dat sundz more like something a "bimbo" might wite. Thiz are done by inserting some wordz with litle meaning like "like", "um" and "baesicly" into uh, reasonable placiez in sentenciez using, like, natural language pwocesing. Wordz are also mispeled using a curated list of mispelingz and some rulez (like replacing t with t in wordz). De curated list of mispelingz iz also used to basicaly replace wordz with synonymz dat sund more sooo bimbo-ish, or yah nkow, mean something slightly absolutely uh, diferent (ex: than -> dan). In adition, numberz like fifty, or one dusand, or nine dusand are able to be scwambled and runded.

Some odd sentences will always arise, but this should improve over time.

Installation

There are two versions of the library. One for use with node mispell.node.js, and one for easy inclusion in html mispell.js. If you don't want to build them yourself you can download the latest release or install using npm. (Remember to spell it with only one S or you'll get a different package!)

npm i mispell --save

Usage

The library exposes the following functions:

• bimbofy(String, Float) takes the text you want to transform as the first argument and a number between 0 and 1 (the bimbofactor) to determine how much the text is transformed.
• word_complexity(String) takes a single word and returns a number 0 or greater, the greater the number, the more complex the word.
• scramble(String, Float) takes a text to be transformed and the bimbofactor. The bimbofactor is transformed into a complexity value and used as a cutoff for complexity. Words more complex than the cutoff value will be scrambled.

Node:

// const mispell = require('./mispell.node.js') // if using the file directly
const mispell = require('mispell') // if installed via npm
mispell.bimbofy("Hello", 0.75)

or

// TODO: Shouldn't have to specify dist
import { mispell } from "mispell/dist/mispell.node"

Html + JS, just add a script tag in the html file to load the library before any JS is going to use it:

<script src="/scripts/mispell.js"></script>

Then use it in your JS:

mispell.bimbofy("Hello", 0.75)

Development

Build the project using npm run build. When developing it is usually necessary to try things out for yourself, this is done in the devStart.js file. Run this file with npm run devStart.

Resources

The files res/aspell.dat, res/birkbeck.dat and res/wikipedia.dat contain common misspellings in the following format: Each correct word is preceded by a dollar sign and followed by its misspellings, each on one line, without duplicates. (A misspelling might appear more than once in the corpus, but only as a misspelling of different words.) Where the spelling or misspelling contained a space, this has been replaced by an underscore (a_lot, Christ_mas). While most of the misspellings are non-words, there are also some real-word errors, such as "omitted" for "admitted". For more information, see the page they were downloaded from.

The file res/bimbo-dict.json is a custom-built dictionary using the same format as dictionaries are read into internally. Each key is a correctly spelled word and may point to either a list of misspellings, or a list of objects where each object represents a misspelled word. These objects has a property spelling containing the misspelled word and can optionally have a weight representing how likely this spelling is to come up relative to other misspellings. If the first of these objects instead has a property synonym it will use the same misspellings as another key (listed as the value of the synonym property).

All dictionaries should use lowercase letters only. Capitalization is always preserved from the text being bimbofied.

The file res/20k-common-words.txt is from https://github.com/first20hours/google-10000-english and contains the 20k most common english words in order of how common they are.

Natural Language Processing

Natural language processing is done using the Compromise library. See it's api documentation and match syntax.