bm25f 0.2.3 → 0.2.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b00f60fdde35d0565f169e9486d2935e5ddcef9bbd8084d60d9228364c9ebed7
-  data.tar.gz: 7249983dda5a101fb1b2ddb646414b58776ae9a29feb4f563707e9a654d45094
+  metadata.gz: 7c7af39f073c1fc55d42199f9c882caa5adf672d5c94c715fe379c58cd4a4a19
+  data.tar.gz: 1ebb197c89da4b70780d3631658a1bb3d0ce5c244abe8b5875c41ee1632f63a3
 SHA512:
-  metadata.gz: 16231fa9ba99f19cff2a6e9e801b734b53effa584da0bc8fcc80000774bbe61c75f1994bfbdc39b3cdac8d427b5049ff56dd1dbcc669eed3b836d910e45f5547
-  data.tar.gz: eea02e6dc1989ff69baa6097997da274149083a87cbbc7355101e613103f09fa52886718422cf68b9cb9afadb2e4ed9ce54be8117cba536a99647db01acb24f0
+  metadata.gz: 816270030afd15716f8b89f4c74d5c11dbc32f169c7a10a1eca931782e21e320017aa95bab66e32c6b500d407922e289edce8db396c956a62c245871744bb79b
+  data.tar.gz: 0ef709327eaa9c5b14e1132bafc199c4d66759b509c85d8bf3dcb23ba5b8dbe526b698c0ab1051486741fa3a3d4d21d073f8d4e3d12ced9128ec16fedd6a39e4

data/lib/bm25f.rb

@@ -88,7 +88,7 @@ class BM25F
   def calculate_idf
     idf = {}
     @field_weights.each_key do |field|
-      field_doc_count = @documents.count { |doc| !doc[field].empty? }
+      field_doc_count = @documents.count { |doc| !doc[field]&.empty? }
       idf[field] = Math.log((@total_docs - field_doc_count + 0.5) / (field_doc_count + 0.5) + 1.0)
     end
     idf
@@ -114,7 +114,10 @@ class BM25F
         tf = field_term_frequency(field, term, doc_id)
         idf = @idf[field]
         field_length_norm = field_length_norm(field, doc_id)
-        doc_score += @field_weights[field] * ((tf * (@term_freq_weight + 1)) / (tf + @term_freq_weight * field_length_norm) * idf)
+        val = @field_weights[field] * ((tf * (@term_freq_weight + 1)) / (tf + @term_freq_weight * field_length_norm) * idf)
+        val = 0 if val.nan?
+
+        doc_score += val
       end
     end
     doc_score
@@ -127,7 +130,10 @@ class BM25F
   # @param doc_id [Integer] The document ID.
   # @return [Integer] The term frequency.
   def field_term_frequency(field, term, doc_id)
-    @documents[doc_id][field].scan(term).count
+    val = @documents[doc_id][field]
+    return 0 if val.nil?
+
+    val.scan(term).count
   end
 
   # Calculates the field length normalization factor of a document.
@@ -136,6 +142,9 @@ class BM25F
   # @param doc_id [Integer] The document ID.
   # @return [Float] The field length normalization factor.
   def field_length_norm(field, doc_id)
-    1.0 - @doc_length_weight + @doc_length_weight * (@doc_lengths[doc_id][field] / @avg_doc_length)
+    val = @doc_lengths[doc_id][field]
+    return 0 if val.nil?
+
+    1.0 - @doc_length_weight + @doc_length_weight * (val / @avg_doc_length)
   end
 end

data/test/test_bm25f.rb

@@ -6,7 +6,7 @@ class BM25FTest < Minitest::Test
     @bm25f = BM25F.new
     @documents = [
       { title: 'hello world', content: 'foo bar baz' },
-      { title: 'foo bar', content: 'goodbye, world!' }
+      { title: 'foo bar', content: 'goodbye, world!', test: nil }
     ]
   end
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: bm25f
 version: !ruby/object:Gem::Version
-  version: 0.2.3
+  version: 0.2.4
 platform: ruby
 authors:
 - catflip