annoy-rb provides Ruby bindings for the Annoy (Approximate Nearest Neighbors Oh Yeah).

Add this line to your application's Gemfile:

gem 'annoy-rb'

And then execute:

$ bundle install

Or install it yourself as:

$ gem install annoy-rb

Note: annoy-rb does not require the installation of another external library. In addition, annoy-rb does not give any optimization options when building native extensions. If necessary, add optimization options yourself during installation, as follows;

$ bundle config --local build.annoy-rb "--with-cxxflags=-march=native"
$ bundle install

Or:

$ gem install annoy-rb -- --with-cxxflags=-march=native

• annoy-rb API Documentation

require 'annoy'

f = 40 # length of item vector that will be indexed.
t = Annoy::AnnoyIndex.new(n_features: f, metric: 'angular')

1000.times do |i|
  v = Array.new(f) { rand }
  t.add_item(i, v)
end

t.build(10) # 10 trees.
t.save('test.ann')

u = Annoy::AnnoyIndex.new(n_features: f, metric: 'angular')
u.load('test.ann')
p u.get_nns_by_item(0, 100) # will find the 100 nearest neighbors.

With the default argument, annoy-rb uses double precision floating point type for the data type of vector element. On the other hand, the Python bindings of Annoy use single precision floating point type. If you want to load a search index created with the Python bindings, specify 'float32' to the dtype argument.

require 'annoy'

f = 40
t = Annoy::AnnoyIndex.new(n_features: f, metric: 'angular', dtype: 'float32')
t.load('index_with_python_bindings.ann')

The gem is available as open source under the terms of the Apache-2.0 License.

Bug reports and pull requests are welcome on GitHub at https://github.com/yoshoku/annoy-rb. This project is intended to be a safe, welcoming space for collaboration, and contributors are expected to adhere to the Contributor Covenant code of conduct.