Fuzzinator

Fuzzinator is a fuzzy search package that uses vectors in order to score searches. It is highly tuneable.

Tuning (configuration)

Scalars

• chars
  • This is a list of characters that are recognized and will have a unique scalar value.
  • The closer two letters are in the array, the closer their scalar values will be together.
  • Capitals do not need to be included in this list, and including them will give them a unique scalar value at least 1 away their lower case equivalent.
  • Characters not in the char list that are encountered when converting a string to a vector are added to the end of the char list, and the scalar is assessed accordingly.
• capitalDistance
  • The difference in scalar value capital letters (that are not included in the chars list) will have compared to their lower case equivalents. Between 0 and 1 (exclusive)

Scoring

• Multiplier
  • initial
    • The initial multiplier value applied to each component of the vector. Typically starts at 1
  • reducer
    • The amount to divide the multiplier by each subsequent value added to the vector
  • reductionDelimeter
    • The character that triggers a reduction in the multiplier (empty string "" means multiplier is reduced after each character)
  • resetReducer
    • The amount to divide the inital multiplier by when it is restored
  • resetDelimeter
    • The character that triggers the multiplier to be reset
• Formula
  • The formula for computing distance. Either "l1" or "l2"
    • "l1" is the Manhattan formula
    • "l2" is the Squared Euclidian formula

Usage

Fuzzinator Class

To get started, create a new instance of the Fuzzinator class. To tune the fuzzinator, pass a configuration object during instantiation.

Fuzzinator instances have a method rank attached to them. This method takes a "root" value (string), and a "space" value (string[]). The root value is compared to all of the values in the space. It returns an array of VString objects, which is an extension of the string class, which extra fields distance, vector, and score attached to them. The score field is a value ranging from 0 to 1, with 1 representing an identical match.

The rank method can also take a distance threshold. This is the minimum distance a string must be from the root string.

Rank

import Fuzzinator from "fuzzinator";

const fz = new Fuzzinator({
    scalars : {
        chars: [' ','a','e','i','o','u','b','c','d','e', ...],
        capitalDistance: .001,
    },
    scoring : {
        multiplier : {
            initial : 1,
            reducer : 1.2,
            reductionDelimeter: "",
            resetDelimeter: " ",
            resetReducer: 1.1
        },
        formula: "l2"
    }
});

const root = "Hello World"; 
const space = [
    "hello world",
    "HELLO WORLD",
    "Photos printed?",
    "Bogos binted"
];

const rankings = fz.rank({
    root,
    space
});

const filteredRankings = fz.rank({
    root,
    space,
    threshold: 10
})

Other Methods

computeVector

The computeVector method can be used to get the raw Vector for a string.

import Fuzzinator, {Vector} from "fuzzinator"; 

const fz = new Fuzzinator(); 

const myVector:Vector = fz.computeVector("George Goosington");

computeDistance

The computeDistance method can be used to calculate the distance between two Vectors or strings.

import Fuzzinator, {Vector} from "fuzzinator"; 

const fz = new Fuzzinator(); 

const v1:Vector = fz.computeVector("George Goosington");
const v2:Vector = fz.computeVector("Thomas Skunkerson");

const distance:number = fz.computeDistance(v1, v2);