### Complete Search Implementation Example with BM25F in Ruby

Source: https://context7.com/catflip/bm25f-ruby/llms.txt

A comprehensive example demonstrating a full search system implementation using BM25F. It covers initializing the model, fitting it with documents and field weights, and then scoring queries to rank results.

```ruby
require 'bm25f'

# Sample document collection
documents = [
  {
    url: 'https://wikimedia.org',
    title: 'Wikimedia',
    content: 'Wikimedia is a global movement whose mission is to bring free educational content to the world. Through various projects, chapters and the support structure.'
  },
  {
    url: 'https://twitter.com/Wikipedia',
    title: 'Wikipedia (@Wikipedia) · Twitter',
    content: nil  # Handles nil content gracefully
  },
  {
    url: 'https://play.google.com/store/apps/details',
    title: 'Wikipedia - Apps on Google Play',
    content: 'The best Wikipedia experience on your Mobile device. Ad-free and free of charge, forever.'
  },
  {
    url: 'https://www.wikipedia.org',
    title: 'Wikipedia',
    content: 'Wikipedia is a free online encyclopedia, created and edited by volunteers around the world and hosted by the Wikimedia Foundation.'
  }
]

# Initialize and fit model
bm25f = BM25F.new(term_freq_weight: 1.33, doc_length_weight: 0.8)
bm25f.fit(documents, { url: 0.3, title: 1.5, content: 1.0 })
```

--------------------------------

### Install bm25f Ruby Gem

Source: https://github.com/catflip/bm25f-ruby/blob/main/README.md

Instructions for installing the bm25f gem using either Bundler for Gemfile integration or the gem command-line tool.

```ruby
gem 'bm25f', '~> 0.2.0'
```

```shell
$ gem install bm25f
```

--------------------------------

### Use BM25F Ranking Algorithm in Ruby

Source: https://github.com/catflip/bm25f-ruby/blob/main/README.md

Example demonstrating the usage of the bm25f-ruby gem. It shows how to initialize the BM25F model, load document data with custom weights, and calculate similarity scores for a given query. The output displays the URL and its corresponding score.

```ruby
require 'bm25f'

# Example data
data = [
  { url: 'https://example.site', title: 'Example Site', description: 'Lorem ipsum dolor sit amet.' },
  { url: 'https://test.website', title: 'Test Website', description: 'A site for testing stuff.' }
]

query = 'test site'

# Create a new BM25F model
bm25f = BM25F.new

# Load the document data in the model using custom weights
bm25f.fit data, { url: 0.5, title: 1, description: 0.8 }

# Calculate the score of the data
score = bm25f.score query

score.each do |k, v|
  # Print out the URL, the query and the score
  puts "#{data[k][:url]} is similar to '#{query}' by #{v}"
end
```

--------------------------------

### Initialize BM25F Model in Ruby

Source: https://context7.com/catflip/bm25f-ruby/llms.txt

Creates a new BM25F model instance. You can initialize with default parameters or provide custom values for `term_freq_weight` (k1) and `doc_length_weight` (b) to fine-tune the scoring.

```ruby
require 'bm25f'

# Initialize with default parameters (term_freq_weight: 1.33, doc_length_weight: 0.8)
bm25f = BM25F.new

# Initialize with custom parameters for fine-tuning
bm25f_custom = BM25F.new(term_freq_weight: 1.5, doc_length_weight: 0.75)
```

--------------------------------

### Train BM25F Model on Documents in Ruby

Source: https://context7.com/catflip/bm25f-ruby/llms.txt

Fits the BM25F model to a collection of documents. Documents are provided as an array of hashes, where each hash represents a document with field-value pairs. You can optionally specify field weights to prioritize certain fields during ranking.

```ruby
require 'bm25f'

bm25f = BM25F.new

# Document collection - each document is a hash with multiple fields
documents = [
  {
    url: 'https://wikimedia.org',
    title: 'Wikimedia',
    content: 'Wikimedia is a global movement whose mission is to bring free educational content to the world.'
  },
  {
    url: 'https://www.wikipedia.org',
    title: 'Wikipedia',
    content: 'Wikipedia is a free online encyclopedia, created and edited by volunteers around the world.'
  },
  {
    url: 'https://play.google.com/store/apps/details',
    title: 'Wikipedia - Apps on Google Play',
    content: 'The best Wikipedia experience on your Mobile device. Ad-free and free of charge, forever.'
  }
]

# Fit with default field weights (all fields weighted equally at 1.0)
bm25f.fit(documents)

# Or fit with custom field weights - higher weight = more importance
bm25f.fit(documents, { url: 0.5, title: 1.5, content: 1.0 })
```

--------------------------------

### Ruby BM25F Search and Display Results

Source: https://context7.com/catflip/bm25f-ruby/llms.txt

This snippet defines and executes a search function using BM25F in Ruby. It scores documents against a query, ranks them by relevance, and then iterates through the results to print their scores, titles, and URLs. It assumes the existence of a BM25F object, a documents hash, and the search function definition.

```ruby
def search(bm25f, documents, query, top_n: 3)
  scores = bm25f.score(query)

  ranked = scores.sort_by { |_id, score| -score }
                 .first(top_n)
                 .reject { |_id, score| score <= 0 }

  ranked.map do |doc_id, score|
    {
      score: score.round(4),
      title: documents[doc_id][:title],
      url: documents[doc_id][:url]
    }
  end
end

# Execute search
results = search(bm25f, documents, 'wikipedia encyclopedia')

results.each do |result|
  puts "[#{result[:score]}] #{result[:title]}"
  puts "  #{result[:url]}"
end
```

--------------------------------

### Calculate Document Scores with BM25F in Ruby

Source: https://context7.com/catflip/bm25f-ruby/llms.txt

Calculates relevance scores for documents against a query. The `score` method returns a hash mapping document indices to their BM25F scores. These scores can be used to rank documents by relevance.

```ruby
require 'bm25f'

bm25f = BM25F.new

documents = [
  { url: 'https://example.site', title: 'Example Site', description: 'Lorem ipsum dolor sit amet.' },
  { url: 'https://test.website', title: 'Test Website', description: 'A site for testing stuff.' },
  { url: 'https://sample.page', title: 'Sample Page', description: 'Another sample page for demonstration.' }
]

# Fit model with field weights
bm25f.fit(documents, { url: 0.5, title: 1.0, description: 0.8 })

# Score documents against query
query = 'test site'
scores = bm25f.score(query)
# => { 0 => 0.234, 1 => 1.567, 2 => 0.089 }

# Sort documents by relevance score (highest first)
ranked_results = scores.sort_by { |_doc_id, score| -score }

ranked_results.each do |doc_id, score|
  puts "Score: #{score.round(3)} - #{documents[doc_id][:title]}"
end
# Output:
# Score: 1.567 - Test Website
# Score: 0.234 - Example Site
# Score: 0.089 - Sample Page
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.