Namenormalizer NPM

namenormalizer

Nodejs class for normalizing retirement plan names for easier voice searching.

Installation

npm install -save namenormalizer

Usage

Usually, this package will be used in a loop to process a list of benefit plan names. This example shows how to instantiate the class and call it one time.

var namenormalizer = require('namenormalizer');
var originalNames  = [
    "1 & 1 INTERNET INC.",
    "1-800 CONTACTS, INC.",
    "2ND WIND EXERCISE EQUIPMENT, INC.",
    "A & S SERVICES GROUP, LLC",
    "AAA LIFE INSURANCE COMPANY",
    "BRGC INTL"
];

for (var i = 0; i < originalNames.length; i++)
{
    console.log(namenormalizer.normalize(originalNames[i]));
}

This will output the following:

ONE AND ONE INTERNET
ONE EIGHT ZERO ZERO CONTACTS
SECOND WIND EXERCISE EQUIPMENT
A AND S SERVICES
TRIPLE A LIFE INSURANCE
B.R.G.C. INTERNATIONAL

Notes

I can't imagine that this package has any general use. It is specifically designed to take employee benefit program sponsor names and convert the odd abbreviations, stop words, and noise text into something that is easy to search for using voice input. In particular, it has been developed for my use in implementing a custom "Alexa" skill using Amzaon's Echo Voice Services.

To Do

First, The current implementation converts "ABC" to "A.B.C.", which is easier for the Amazon Echo voice recognizer to recognize. It also converts acronyms comprised solely of consonants into the same form. For example, "BGC" is translated to "B.G.C." for easier recognition. However, "UAW" is not converted to an acronym because it contains vowels. I didn't want to "hardcode" a bunch of acronyms, as I did for "ABC", yet my consonant-only approach misses many aconyms. For these reasons I may need to connect this to some spell checker and where we can't find the spelling, then convert the letters to an acronym. It would probably be a good idea to limit that to three-letter or four-letter acronyms. There are many proper names that would not be found in a spelling dictionary and which should not be converted to acronyms, such as "ACXIOM" and "ZYNGA".