bm25f 0.2.2 → 0.2.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: bc84b7827c64bd77694f548ac479113906f4a5173b50fab2c479079beec3bc41
-  data.tar.gz: 8adf6005477365e0b827e93b67ae8b4bbc99c85db73e1b274005ce043659d690
+  metadata.gz: 7c7af39f073c1fc55d42199f9c882caa5adf672d5c94c715fe379c58cd4a4a19
+  data.tar.gz: 1ebb197c89da4b70780d3631658a1bb3d0ce5c244abe8b5875c41ee1632f63a3
 SHA512:
-  metadata.gz: 17f1f1e79d4fa265610c85dbc52ae949de236252d96796beca5a08c80c4035947b56dc04607ca00aa5e9e2eaac41cb9f5a645f78415c66c1b91aaa5a77c25e60
-  data.tar.gz: fbc526d3080a2672b41da06b78c7df363138bbc62df91bc8c065133ec2e807432c19ea108fdc14fb3bb2852973bfefdb18d0a0f98a4d4a7742eac23a1d7afdbb
+  metadata.gz: 816270030afd15716f8b89f4c74d5c11dbc32f169c7a10a1eca931782e21e320017aa95bab66e32c6b500d407922e289edce8db396c956a62c245871744bb79b
+  data.tar.gz: 0ef709327eaa9c5b14e1132bafc199c4d66759b509c85d8bf3dcb23ba5b8dbe526b698c0ab1051486741fa3a3d4d21d073f8d4e3d12ced9128ec16fedd6a39e4

data/lib/bm25f.rb

@@ -77,7 +77,7 @@ class BM25F
   def calculate_document_lengths(documents)
     doc_lengths = {}
     documents.each_with_index do |doc, i|
-      doc_lengths[i] = doc.transform_values { |v| v.nil ? 0 : v.length }
+      doc_lengths[i] = doc.transform_values { |v| v.nil? ? 0 : v.length }
     end
     doc_lengths
   end
@@ -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: 0.2.2
+  version: 0.2.4
 platform: ruby
 authors:
 - catflip