Project

rrf

0.0
The project is in a healthy, maintained state
This gem provides an implementation of Reciprocal Rank Fusion (RRF) to merge results from different search engines. Initially supporting Active Record and Elasticsearch.
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
 Dependencies

Development

 Project Readme

RRF

RRF (Reciprocal Rank Fusion) is a Ruby gem that provides an implementation of the Reciprocal Rank Fusion (RRF) algorithm to merge and rank results from different search engines, specifically supporting ActiveRecord and Searchkick.

Installation

Add this line to your application's Gemfile:

gem 'rrf'

And then execute:

bundle install

Or install it yourself as:

gem install rrf

Usage

To use RRF, include the module in your ActiveRecord model and configure it as needed.

Example

Let’s assume you have a model named Chunk that uses both ActiveRecord (with pgvector) and Searchkick:

class Chunk < ApplicationRecord
  include Searchkick
  include RRF::Model

  # ActiveRecord vector search using pgvector
  # Assuming `embedding768` is a column of type `vector` (pgvector)

  # Define any additional logic or methods as needed
end

Performing Searches and Fusing Results

You can perform searches using ActiveRecord and Searchkick, and then fuse the results using the fuse method (the following relies on the neighbor gem and the red-candle gem):

# Configure the constant value if needed
RRF.configure do |config|
  config.rank_bias = 70 # User-defined value
end

# Perform Searchkick search
es_result = Chunk.search("alpaca", load: false, limit: 50)

# Perform ActiveRecord nearest neighbor search
query_embedding = Candle::Model.new.embedding("alpaca") # You'd want to cache the model in memory
ar_result = Chunk.all.nearest_neighbors(:embedding768, query_embedding, distance: :cosine).limit(50)

# Fuse the results and limit to 10
result = Chunk.fuse(ar_result, es_result, limit: 10)

# `result` now contains the top 10 fused search results

Configuration

You can configure the rank_bias value used in the RRF algorithm:

RRF.configure do |config|
  config.rank_bias = 70 # Default is 60
end

Development

After checking out the repo, run

bundle install

Then, run

rake spec

to run the tests.

To install this gem onto your local machine, run

bundle exec rake install

To release a new version, update the version number in version.rb, and then run

bundle exec rake release

which will create a git tag for the version, push git commits and tags, and push the .gem file to rubygems.org.

To use this locally during development, I like to include it whatever project I'm working by adding the following to the Gemfile of the other project

gem "rrf", path: '/what/ever/path/you/choose/rrf'

Then bundle and it will use your development copy.

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/assaydepot/rrf.

License

The gem is available as open source under the terms of the MIT License.