string_similarity_comparator

Return a json aggregating different algorithms to compare strings. This can help compare results from different algorithms and better choose the most appropriate depending of the objective.

Algorithms

• Cosine
• Levenshtein distance
• Levenshtein
• Trigram
• Jaro Winkler
• Jaro
• Hamming

Installation

First thing first, add this line to your Gemfile

gem 'string_similarity_comparator'

Then, run

$ bundle

Or install it yourself as:

$ gem install 'string_similarity_comparator'

Usage

1. Overview as gem

You can test it in console right away.

>> require 'string_similarity_comparator'
true

>> StringSimilarityComparator::Pool.new('foo', 'bar').calculate
{
                  :cosine => 0.0,
             :levenshtein => 0.333,
    :levenshtein_distance => 3,
                 :trigram => 0.0,
            :jaro_winkler => 0.0,
                    :jaro => 0.0,
                 :hamming => 3
}

To compare two words, 'foo' and 'bar' for example, call

>> StringSimilarityComparator::Pool.new('foo', 'bar').calculate

2. Overview as sinatra API

$ git clone [email protected]:coralieco/string_similarity_comparator.git

Then

$ cd string_similarity_comparator
$ bundle
$ ruby lib/string_similarity_comparator/app.rb

Go on the browser, usually on localhost:4567

Then use one of the endpoint:

1. with the form: http://localhost:4567/

2. with the url directly: http://localhost:4567/api/v1/string_similarity?string_a=foo&string_b=bar

Article

I wrote an article about what we do at Appaloosa Store to customize application recommendation using String Similarity Algorithms: String Similarity Algorithms Compared

It compares some of these algorithms and explain why I chose to use the Jaro-Winkler algorithm in Appaloosa use-case.

License

Using gem work of: See the LICENSE file in the source.

JaroWinkler: Copyright (c) 2014 Jian Weihang

amatch: Copyright [2017] [Florian Frank]

string-similarity: Copyright (c) 2015 Manuel Hutter

Trigram: Copyright (c) 2014 milk1000cc