From 090cb156acbfcaec460415180a5d4af37b1c515f Mon Sep 17 00:00:00 2001
From: yun-zhi-ztl <15071461069@163.com>
Date: Sun, 15 Mar 2026 17:09:46 +0800
Subject: [PATCH] tune(search): improve semantic reranking quality

---
 .../search/HashingSearchEmbeddingService.java | 23 +++++++++++++++----
 .../PostgresFullTextIndexService.java         |  1 +
 .../HashingSearchEmbeddingServiceTest.java    | 11 +++++++++
 3 files changed, 31 insertions(+), 4 deletions(-)

diff --git a/server/skillhub-search/src/main/java/com/iflytek/skillhub/search/HashingSearchEmbeddingService.java b/server/skillhub-search/src/main/java/com/iflytek/skillhub/search/HashingSearchEmbeddingService.java
index 1ed4a72c..b92efe24 100644
--- a/server/skillhub-search/src/main/java/com/iflytek/skillhub/search/HashingSearchEmbeddingService.java
+++ b/server/skillhub-search/src/main/java/com/iflytek/skillhub/search/HashingSearchEmbeddingService.java
@@ -10,6 +10,7 @@ import org.springframework.stereotype.Service;
 public class HashingSearchEmbeddingService implements SearchEmbeddingService {
     private static final Pattern TOKEN_SPLITTER = Pattern.compile("[^\\p{L}\\p{N}_]+");
     private static final int DIMENSIONS = 64;
+    private static final double NGRAM_WEIGHT = 0.35D;
 
     @Override
     public String embed(String text) {
@@ -47,16 +48,30 @@ public class HashingSearchEmbeddingService implements SearchEmbeddingService {
                 .map(String::trim)
                 .filter(token -> !token.isBlank())
                 .forEach(token -> {
-                    int hash = token.hashCode();
-                    int index = Math.floorMod(hash, DIMENSIONS);
-                    double weight = 1D + Math.min(token.length(), 12) / 12D;
-                    vector[index] += weight;
+                    addTokenWeight(vector, token, 1D + Math.min(token.length(), 12) / 12D);
+                    addCharacterNgrams(vector, token);
                 });
 
         normalize(vector);
         return vector;
     }
 
+    private void addTokenWeight(double[] vector, String token, double weight) {
+        int hash = token.hashCode();
+        int index = Math.floorMod(hash, DIMENSIONS);
+        vector[index] += weight;
+    }
+
+    private void addCharacterNgrams(double[] vector, String token) {
+        if (token.length() < 3) {
+            return;
+        }
+        for (int i = 0; i <= token.length() - 3; i++) {
+            String trigram = token.substring(i, i + 3);
+            addTokenWeight(vector, trigram, NGRAM_WEIGHT);
+        }
+    }
+
     private double[] parseVector(String serializedVector) {
         String[] parts = serializedVector.split(",");
         double[] vector = new double[parts.length];
diff --git a/server/skillhub-search/src/main/java/com/iflytek/skillhub/search/postgres/PostgresFullTextIndexService.java b/server/skillhub-search/src/main/java/com/iflytek/skillhub/search/postgres/PostgresFullTextIndexService.java
index 7db8bcca..b856b82a 100644
--- a/server/skillhub-search/src/main/java/com/iflytek/skillhub/search/postgres/PostgresFullTextIndexService.java
+++ b/server/skillhub-search/src/main/java/com/iflytek/skillhub/search/postgres/PostgresFullTextIndexService.java
@@ -75,6 +75,7 @@ public class PostgresFullTextIndexService implements SearchIndexService {
 
     private String buildSemanticVector(SkillSearchDocument document) {
         return searchEmbeddingService.embed(String.join("\n",
+                safe(document.title()),
                 safe(document.title()),
                 safe(document.summary()),
                 safe(document.keywords()),
diff --git a/server/skillhub-search/src/test/java/com/iflytek/skillhub/search/HashingSearchEmbeddingServiceTest.java b/server/skillhub-search/src/test/java/com/iflytek/skillhub/search/HashingSearchEmbeddingServiceTest.java
index d916cc8d..c0412a16 100644
--- a/server/skillhub-search/src/test/java/com/iflytek/skillhub/search/HashingSearchEmbeddingServiceTest.java
+++ b/server/skillhub-search/src/test/java/com/iflytek/skillhub/search/HashingSearchEmbeddingServiceTest.java
@@ -26,4 +26,15 @@ class HashingSearchEmbeddingServiceTest {
 
         assertThat(relevant).isGreaterThan(noisy);
     }
+
+    @Test
+    void similarityShouldHandleSingularAndPluralForms() {
+        String pluralVector = service.embed("build strong habits with daily practice");
+        String unrelatedVector = service.embed("research company profiles on the web");
+
+        double pluralMatch = service.similarity("habit", pluralVector);
+        double unrelatedMatch = service.similarity("habit", unrelatedVector);
+
+        assertThat(pluralMatch).isGreaterThan(unrelatedMatch);
+    }
 }