Merge pull request #31813 from vespa-engine/toregge/add-document-freq…

…uency-to-query-items Add document frequency to query items.
vespa-engine · Jul 2, 2024 · d1b3b4c · d1b3b4c
2 parents e0ff928 + c700bab
commit d1b3b4c
Show file tree

Hide file tree

Showing 8 changed files with 91 additions and 3 deletions.
diff --git a/container-search/abi-spec.json b/container-search/abi-spec.json
@@ -480,10 +480,30 @@
       "public void setExplicitSignificance(boolean)",
       "public boolean hasExplicitSignificance()",
       "public double getSignificance()",
+      "public void setDocumentFrequency(com.yahoo.prelude.query.DocumentFrequency)",
+      "public java.util.Optional getDocumentFrequency()",
       "public boolean hasUniqueID()"
     ],
     "fields" : [ ]
   },
+  "com.yahoo.prelude.query.DocumentFrequency" : {
+    "superClass" : "java.lang.Record",
+    "interfaces" : [ ],
+    "attributes" : [
+      "public",
+      "final",
+      "record"
+    ],
+    "methods" : [
+      "public void <init>(long, long)",
+      "public final java.lang.String toString()",
+      "public final int hashCode()",
+      "public final boolean equals(java.lang.Object)",
+      "public long frequency()",
+      "public long count()"
+    ],
+    "fields" : [ ]
+  },
   "com.yahoo.prelude.query.DotProductItem" : {
     "superClass" : "com.yahoo.prelude.query.WeightedSetItem",
     "interfaces" : [ ],
@@ -874,7 +894,8 @@
       "protected com.yahoo.prelude.query.Item connectedBacklink",
       "protected double connectivity",
       "protected double significance",
-      "protected boolean explicitSignificance"
+      "protected boolean explicitSignificance",
+      "protected com.yahoo.prelude.query.DocumentFrequency documentFrequency"
     ]
   },
   "com.yahoo.prelude.query.ItemHelper" : {
@@ -1578,6 +1599,8 @@
       "public void setExplicitSignificance(boolean)",
       "public boolean hasExplicitSignificance()",
       "public double getSignificance()",
+      "public void setDocumentFrequency(com.yahoo.prelude.query.DocumentFrequency)",
+      "public java.util.Optional getDocumentFrequency()",
       "public boolean hasUniqueID()"
     ],
     "fields" : [ ]
@@ -1679,7 +1702,9 @@
       "public abstract void setSignificance(double)",
       "public abstract boolean hasExplicitSignificance()",
       "public abstract void setExplicitSignificance(boolean)",
-      "public abstract double getSignificance()"
+      "public abstract double getSignificance()",
+      "public abstract void setDocumentFrequency(com.yahoo.prelude.query.DocumentFrequency)",
+      "public abstract java.util.Optional getDocumentFrequency()"
     ],
     "fields" : [ ]
   },
@@ -1703,6 +1728,8 @@
       "public void setExplicitSignificance(boolean)",
       "public boolean hasExplicitSignificance()",
       "public double getSignificance()",
+      "public void setDocumentFrequency(com.yahoo.prelude.query.DocumentFrequency)",
+      "public java.util.Optional getDocumentFrequency()",
       "public boolean hasUniqueID()"
     ],
     "fields" : [ ]

diff --git a/container-search/src/main/java/com/yahoo/prelude/query/CompositeTaggableItem.java b/container-search/src/main/java/com/yahoo/prelude/query/CompositeTaggableItem.java
@@ -1,6 +1,8 @@
 // Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
 package com.yahoo.prelude.query;
 
+import java.util.Optional;
+
 /**
  * Common implementation for Item classes implementing the TaggableItem interface.
  * Note that this file exists in 3 copies that should be kept in sync:
@@ -68,6 +70,12 @@ public double getSignificance() {
         return significance;
     }
 
+    @Override
+    public void setDocumentFrequency(DocumentFrequency documentFrequency) { this.documentFrequency = documentFrequency; }
+
+    @Override
+    public Optional<DocumentFrequency> getDocumentFrequency() { return Optional.ofNullable(documentFrequency); }
+
     //Change access privilege from protected to public.
     public boolean hasUniqueID() {
         return super.hasUniqueID();

diff --git a/container-search/src/main/java/com/yahoo/prelude/query/DocumentFrequency.java b/container-search/src/main/java/com/yahoo/prelude/query/DocumentFrequency.java
@@ -0,0 +1,16 @@
+// Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+package com.yahoo.prelude.query;
+
+import com.yahoo.api.annotations.Beta;
+
+/**
+ * The expected number of documents matching an item given a corpus of
+ * multiple documents. This is the raw data used to calculate variants
+ * of idf, used as significance.
+ *
+ * @param frequency The number of documents in which an item occurs
+ * @param count     The total number of documents in the corpus
+ */
+@Beta
+public record DocumentFrequency(long frequency, long count) {
+}
diff --git a/container-search/src/main/java/com/yahoo/prelude/query/Item.java b/container-search/src/main/java/com/yahoo/prelude/query/Item.java
@@ -124,6 +124,8 @@ public enum ItemCreator {
     protected double significance = 0;
     protected boolean explicitSignificance = false;
 
+    protected DocumentFrequency documentFrequency = null;
+
     /** Whether this item is eligible for change by query rewriters (false) or should be kept as-is (true) */
     private boolean isProtected;
 
@@ -495,6 +497,8 @@ public void disclose(Discloser discloser) {
             discloser.addProperty("usePositionData", usePositionData);
         if (explicitSignificance)
             discloser.addProperty("significance", significance);
+        if (documentFrequency != null)
+            discloser.addProperty("documentFrequency", documentFrequency);
         if (weight != 100)
             discloser.addProperty("weight", weight);
         if (label != null)

diff --git a/container-search/src/main/java/com/yahoo/prelude/query/SimpleTaggableItem.java b/container-search/src/main/java/com/yahoo/prelude/query/SimpleTaggableItem.java
@@ -1,6 +1,8 @@
 // Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
 package com.yahoo.prelude.query;
 
+import java.util.Optional;
+
 /**
  * Common implementation for Item classes implementing the TaggableItem interface.
  * Note that this file exist in 3 copies that should be kept in sync:
@@ -68,6 +70,12 @@ public double getSignificance() {
         return significance;
     }
 
+    @Override
+    public void setDocumentFrequency(DocumentFrequency documentFrequency) { this.documentFrequency = documentFrequency; }
+
+    @Override
+    public Optional<DocumentFrequency> getDocumentFrequency() { return Optional.ofNullable(documentFrequency); }
+
     //Change access privilege from protected to public.
     public boolean hasUniqueID() {
         return super.hasUniqueID();

diff --git a/container-search/src/main/java/com/yahoo/prelude/query/TaggableItem.java b/container-search/src/main/java/com/yahoo/prelude/query/TaggableItem.java
@@ -1,5 +1,8 @@
 // Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
 package com.yahoo.prelude.query;
+import com.yahoo.api.annotations.Beta;
+
+import java.util.Optional;
 
 /**
  * An interface used for anything which may be addressed using an external,
@@ -44,4 +47,9 @@ public interface TaggableItem {
     void setExplicitSignificance(boolean significance);
     double getSignificance();
 
+    @Beta
+    void setDocumentFrequency(DocumentFrequency documentFrequency);
+
+    @Beta
+    Optional<DocumentFrequency> getDocumentFrequency();
 }
diff --git a/container-search/src/main/java/com/yahoo/prelude/query/TaggableSegmentItem.java b/container-search/src/main/java/com/yahoo/prelude/query/TaggableSegmentItem.java
@@ -1,6 +1,8 @@
 // Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
 package com.yahoo.prelude.query;
 
+import java.util.Optional;
+
 /**
  * Common implementation for Item classes implementing the TaggableItem interface.
  * Note that this file exist in 3 copies that should be kept in sync:
@@ -81,6 +83,12 @@ public double getSignificance() {
         return significance;
     }
 
+    @Override
+    public void setDocumentFrequency(DocumentFrequency documentFrequency) { this.documentFrequency = documentFrequency; }
+
+    @Override
+    public Optional<DocumentFrequency> getDocumentFrequency() { return Optional.ofNullable(documentFrequency); }
+
     //Change access privilege from protected to public.
     @Override
     public boolean hasUniqueID() {

diff --git a/container-search/src/test/java/com/yahoo/prelude/query/TaggableItemsTestCase.java b/container-search/src/test/java/com/yahoo/prelude/query/TaggableItemsTestCase.java
@@ -99,7 +99,7 @@ void requireSimilarAPIs() {
                 .getDeclaredMethods();
         final Method[] simple = SimpleTaggableItem.class.getDeclaredMethods();
         final Method[] segment = TaggableSegmentItem.class.getDeclaredMethods();
-        final int numberOfMethods = 10;
+        final int numberOfMethods = 12;
         assertEquals(numberOfMethods, composite.length);
         assertEquals(numberOfMethods, simple.length);
         assertEquals(numberOfMethods, segment.length);
@@ -152,4 +152,13 @@ final void testSetSignificance() {
         assertTrue(p.hasExplicitSignificance());
     }
 
+    @Test
+    final void testSetDocumentFrequency() {
+        final PhraseSegmentItem p = new PhraseSegmentItem("farmyards", false, false);
+        assertFalse(p.getDocumentFrequency().isPresent());
+        p.setDocumentFrequency(new DocumentFrequency(13, 100));
+        assertTrue(p.getDocumentFrequency().isPresent());
+        assertEquals(new DocumentFrequency(13, 100), p.getDocumentFrequency().get());
+    }
+
 }