Merge pull request #4291 from vespa-engine/toregge/test-handling-of-m…

…issing-interleaved-features-in-bm25-feature Test handling of missing interleaved features in bm25 feature.
vespa-engine · Nov 29, 2024 · de1ed05 · de1ed05
2 parents 61ca9f3 + 593ef89
commit de1ed05
Show file tree

Hide file tree

Showing 4 changed files with 36 additions and 20 deletions.
diff --git a/tests/search/bm25_feature/bm25_feature.rb b/tests/search/bm25_feature/bm25_feature.rb
@@ -64,8 +64,8 @@ def test_enable_bm25_feature
     # Average field length for content = 4 ((7 + 3 + 2) / 3).
     # Average field length for contenta = 8 ((14 + 6 + 4) / 3).
     feed_and_wait_for_docs("test", 3, :file => @test_dir + "docs.json")
-    assert_no_bm25_scores
-    assert_no_bm25_array_scores
+    assert_degraded_bm25_scores(3)
+    assert_degraded_bm25_array_scores(3)
 
     redeploy(SearchApp.new.sd("#{@test_dir}1/test.sd"))
     60.times do |i|
@@ -208,24 +208,32 @@ def assert_bm25_array_scores(total_doc_count, avg_field_length)
     assert_scores_for_query("contenta:b&type=all", [score(1, 6, idf(2, total_doc_count), avg_field_length),
                                                     score(1, 14, idf(2, total_doc_count), avg_field_length)], 'contenta')
 
-    assert_scores_for_query("content:a+content:d&type=all", [score(1, 4, idf(3, total_doc_count), avg_field_length) + score(1, 4, idf(2, total_doc_count), avg_field_length),
-                                                             score(3, 14, idf(3, total_doc_count), avg_field_length) + score(1, 14, idf(2, total_doc_count), avg_field_length)], 'content')
+    assert_scores_for_query("contenta:a+contenta:d&type=all", [score(1, 4, idf(3, total_doc_count), avg_field_length) + score(1, 4, idf(2, total_doc_count), avg_field_length),
+                                                             score(3, 14, idf(3, total_doc_count), avg_field_length) + score(1, 14, idf(2, total_doc_count), avg_field_length)], 'contenta')
   end
 
-  def assert_no_bm25_scores
-    assert_scores_for_query("content:a&type=all", [0.0, 0.0, 0.0], 'content')
+  def assert_degraded_bm25_scores(total_doc_count)
+    assert_scores_for_query("content:a&type=all", [idf(3, total_doc_count),
+                                                    idf(3, total_doc_count),
+                                                    idf(3, total_doc_count)], 'content')
 
-    assert_scores_for_query("content:b&type=all", [0.0, 0.0], 'content')
+    assert_scores_for_query("content:b&type=all", [idf(2, total_doc_count),
+                                                   idf(2, total_doc_count)], 'content')
 
-    assert_scores_for_query("content:a+content:d&type=all", [0.0, 0.0], 'content')
+    assert_scores_for_query("content:a+content:d&type=all", [idf(3, total_doc_count) + idf(2, total_doc_count),
+                                                             idf(3, total_doc_count) + idf(2, total_doc_count)], 'content')
   end
 
-  def assert_no_bm25_array_scores
-    assert_scores_for_query("contenta:a&type=all", [0.0, 0.0, 0.0], 'contenta')
+  def assert_degraded_bm25_array_scores(total_doc_count)
+    assert_scores_for_query("contenta:a&type=all", [idf(3, total_doc_count),
+                                                    idf(3, total_doc_count),
+                                                    idf(3, total_doc_count)], 'contenta')
 
-    assert_scores_for_query("contenta:b&type=all", [0.0, 0.0], 'contenta')
+    assert_scores_for_query("contenta:b&type=all", [idf(2, total_doc_count),
+                                                    idf(2, total_doc_count)], 'contenta')
 
-    assert_scores_for_query("content:a+content:d&type=all", [0.0, 0.0], 'content')
+    assert_scores_for_query("contenta:a+contenta:d&type=all", [idf(3, total_doc_count) + idf(2, total_doc_count),
+                                                             idf(3, total_doc_count) + idf(2, total_doc_count)], 'contenta')
   end
 
   def idf(matching_doc_count, total_doc_count = 3)

diff --git a/tests/search/bm25_feature/regen/0/test.sd b/tests/search/bm25_feature/regen/0/test.sd
@@ -14,5 +14,13 @@ search test {
         bm25(content) + bm25(contenta)
       }
     }
+    summary-features {
+      bm25(content)
+      bm25(contenta)
+    }
+    match-features {
+      bm25(content)
+      bm25(contenta)
+    }
   }
 }
diff --git a/tests/search/featurestability/result4.txt b/tests/search/featurestability/result4.txt
@@ -193,18 +193,18 @@ attributeMatch(position_exp_zcurve).queryCompleteness#0.0
 attributeMatch(position_exp_zcurve).significance#0.0
 attributeMatch(position_exp_zcurve).totalWeight#0.0
 attributeMatch(position_exp_zcurve).weight#0.0
-bm25(string_arr)#0.0
+bm25(string_arr)#0.28768207245178085
 bm25(string_exp)#0.28768207245178085
-bm25(string_wset)#0.0
-bm25(url_arr)#0.0
+bm25(string_wset)#0.28768207245178085
+bm25(url_arr)#0.28768207245178085
 bm25(url_arr.fragment)#0.0
 bm25(url_arr.host)#0.0
 bm25(url_arr.hostname)#0.0
 bm25(url_arr.path)#0.0
 bm25(url_arr.port)#0.0
 bm25(url_arr.query)#0.0
 bm25(url_arr.scheme)#0.0
-bm25(url_exp)#0.0
+bm25(url_exp)#0.28768207245178085
 bm25(url_exp.fragment)#0.0
 bm25(url_exp.host)#0.0
 bm25(url_exp.hostname)#0.0

diff --git a/tests/search/featurestability/result5.txt b/tests/search/featurestability/result5.txt
@@ -193,18 +193,18 @@ attributeMatch(position_exp_zcurve).queryCompleteness#0.0
 attributeMatch(position_exp_zcurve).significance#0.0
 attributeMatch(position_exp_zcurve).totalWeight#0.0
 attributeMatch(position_exp_zcurve).weight#0.0
-bm25(string_arr)#0.0
+bm25(string_arr)#0.5753641449035617
 bm25(string_exp)#0.28768207245178085
-bm25(string_wset)#0.0
-bm25(url_arr)#0.0
+bm25(string_wset)#0.5753641449035617
+bm25(url_arr)#0.5753641449035617
 bm25(url_arr.fragment)#0.0
 bm25(url_arr.host)#0.0
 bm25(url_arr.hostname)#0.0
 bm25(url_arr.path)#0.0
 bm25(url_arr.port)#0.0
 bm25(url_arr.query)#0.0
 bm25(url_arr.scheme)#0.0
-bm25(url_exp)#0.0
+bm25(url_exp)#0.5753641449035617
 bm25(url_exp.fragment)#0.0
 bm25(url_exp.host)#0.0
 bm25(url_exp.hostname)#0.0