feat(rag): 知识库检索测试新增混合检索

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/domain/bo/knowledge/KnowledgeFragmentBo.java

@@ -79,4 +79,14 @@ public class KnowledgeFragmentBo extends BaseEntity {
      */
     private String rerankModel;
 
+    /**
+     * 是否启用混合检索
+     */
+    private Boolean enableHybrid;
+
+    /**
+     * 混合检索权重 (0.0-1.0)
+     */
+    private Double hybridAlpha;
+
 }

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/domain/bo/knowledge/KnowledgeInfoBo.java

@@ -92,5 +92,15 @@ public class KnowledgeInfoBo extends BaseEntity {
      */
     private String remark;
 
+    /**
+     * 是否启用混合检索（0 否 1 是）
+     */
+    private Integer enableHybrid;
+
+    /**
+     * 混合检索权重比例 (0.0-1.0)
+     */
+    private Double hybridAlpha;
+
 
 }

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/domain/entity/knowledge/KnowledgeFragment.java

@@ -47,5 +47,10 @@ public class KnowledgeFragment extends BaseEntity {
      */
     private String remark;
 
+    /**
+     * 知识库ID
+     */
+    private Long knowledgeId;
+
 
 }

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/domain/entity/knowledge/KnowledgeInfo.java

@@ -93,5 +93,15 @@ public class KnowledgeInfo extends BaseEntity {
      */
     private String remark;
 
+    /**
+     * 是否启用混合检索（0 否 1 是）
+     */
+    private Integer enableHybrid;
+
+    /**
+     * 混合检索权重比例 (0.0-1.0)
+     */
+    private Double hybridAlpha;
+
 
 }

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/domain/vo/knowledge/KnowledgeFragmentVo.java

@@ -39,7 +39,7 @@ public class KnowledgeFragmentVo implements Serializable {
      * 片段索引下标
      */
     @ExcelProperty(value = "片段索引下标")
-    private Long idx;
+    private Integer idx;
 
     /**
      * 文档内容
@@ -53,5 +53,10 @@ public class KnowledgeFragmentVo implements Serializable {
     @ExcelProperty(value = "备注")
     private String remark;
 
+    /**
+     * 知识库ID
+     */
+    private Long knowledgeId;
+
 
 }

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/domain/vo/knowledge/KnowledgeInfoVo.java

@@ -113,6 +113,19 @@ public class KnowledgeInfoVo implements Serializable {
     @ExcelProperty(value = "备注")
     private String remark;
 
+    /**
+     * 是否启用混合检索（0 否 1 是）
+     */
+    @ExcelProperty(value = "是否启用混合检索", converter = ExcelDictConvert.class)
+    @ExcelDictFormat(readConverterExp = "0=否,1=是")
+    private Integer enableHybrid;
+
+    /**
+     * 混合检索权重比例 (0.0-1.0)
+     */
+    @ExcelProperty(value = "混合检索权重比例")
+    private Double hybridAlpha;
+
     /**
      * 文档数（统计字段，非数据库列）
      */

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/domain/vo/knowledge/KnowledgeRetrievalVo.java

@@ -1,7 +1,9 @@
 package org.ruoyi.domain.vo.knowledge;
 
+import lombok.AllArgsConstructor;
 import lombok.Builder;
 import lombok.Data;
+import lombok.NoArgsConstructor;
 
 import java.io.Serial;
 import java.io.Serializable;
@@ -13,11 +15,33 @@ import java.io.Serializable;
  */
 @Data
 @Builder
+@NoArgsConstructor
+@AllArgsConstructor
 public class KnowledgeRetrievalVo implements Serializable {
 
     @Serial
     private static final long serialVersionUID = 1L;
 
+    /**
+     * 片段ID
+     */
+    private String id;
+
+    /**
+     * 文档ID
+     */
+    private String docId;
+
+    /**
+     * 知识库ID
+     */
+    private Long knowledgeId;
+
+    /**
+     * 分片索引
+     */
+    private Integer idx;
+
     /**
      * 片段内容
      */

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/mapper/knowledge/KnowledgeFragmentMapper.java

@@ -33,4 +33,13 @@ public interface KnowledgeFragmentMapper extends BaseMapperPlus<KnowledgeFragmen
             "GROUP BY doc_id" +
             "</script>")
     List<DocFragmentCountVo> selectFragmentCountByDocIds(@Param("docIds") List<String> docIds);
+    @Select("<script>" +
+            "SELECT id, doc_id AS docId, content, idx, knowledge_id AS knowledgeId " +
+            "FROM knowledge_fragment " +
+            "WHERE knowledge_id = #{knowledgeId} " +
+            "AND MATCH (content) AGAINST (#{query} IN NATURAL LANGUAGE MODE) " +
+            "ORDER BY MATCH (content) AGAINST (#{query} IN NATURAL LANGUAGE MODE) DESC " +
+            "LIMIT #{limit}" +
+            "</script>")
+    List<KnowledgeFragmentVo> searchByKeyword(@Param("knowledgeId") Long knowledgeId, @Param("query") String query, @Param("limit") Integer limit);
 }

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/knowledge/impl/KnowledgeAttachServiceImpl.java

@@ -187,6 +187,7 @@ public class KnowledgeAttachServiceImpl implements IKnowledgeAttachService {
                     String fid = RandomUtil.randomString(10);
                     fids.add(fid);
                     KnowledgeFragment knowledgeFragment = new KnowledgeFragment();
+                    knowledgeFragment.setKnowledgeId(knowledgeId);
                     knowledgeFragment.setDocId(docId);
                     knowledgeFragment.setIdx(i);
                     knowledgeFragment.setContent(chunkList.get(i));

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/knowledge/impl/KnowledgeFragmentServiceImpl.java

@@ -28,6 +28,8 @@ import org.ruoyi.service.vector.VectorStoreService;
 import org.springframework.stereotype.Service;
 import java.util.ArrayList;
 import java.util.stream.Collectors;
+import java.util.*;
+import java.util.concurrent.CompletableFuture;
 
 import java.util.List;
 import java.util.Map;
@@ -180,8 +182,47 @@ public class KnowledgeFragmentServiceImpl implements IKnowledgeFragmentService {
         queryVectorBo.setApiKey(chatModel.getApiKey());
         queryVectorBo.setBaseUrl(chatModel.getApiHost());
 
-        // 3. 执行物理检索
+        // 3. 执行搜索 (向量搜索 + 关键词搜索)
-        List<KnowledgeRetrievalVo> allResults = vectorStoreService.search(queryVectorBo);
+        List<KnowledgeRetrievalVo> allResults;
+        
+        boolean hybridEnabled = Boolean.TRUE.equals(bo.getEnableHybrid()) || 
+                             Integer.valueOf(1).equals(knowledgeInfoVo.getEnableHybrid());
+        
+        if (hybridEnabled) {
+            log.info("执行混合检索: kid={}, query={}", bo.getKnowledgeId(), bo.getQuery());
+            try {
+                // 并行执行向量搜索
+                CompletableFuture<List<KnowledgeRetrievalVo>> vectorFuture = CompletableFuture.supplyAsync(() -> 
+                    vectorStoreService.search(queryVectorBo));
+                
+                // 执行关键词搜索 (MySQL)
+                int limit = bo.getTopK() != null ? bo.getTopK() : 50;
+                List<KnowledgeFragmentVo> keywordFragments = baseMapper.searchByKeyword(bo.getKnowledgeId(), bo.getQuery(), limit);
+                List<KnowledgeRetrievalVo> keywordResults = keywordFragments.stream().map(f -> {
+                    KnowledgeRetrievalVo vo = new KnowledgeRetrievalVo();
+                    vo.setId(f.getId().toString());
+                    vo.setContent(f.getContent());
+                    vo.setDocId(f.getDocId());
+                    vo.setIdx(f.getIdx());
+                    vo.setKnowledgeId(f.getKnowledgeId());
+                    vo.setScore(10.0); // 初始分，后续由 RRF 重新打分
+                    return vo;
+                }).collect(Collectors.toList());
+                
+                List<KnowledgeRetrievalVo> vectorResults = vectorFuture.get();
+                log.info("抽取混合结果成功: Vector命中={}条, Keyword命中={}条", vectorResults.size(), keywordResults.size());
+
+                double alpha = bo.getHybridAlpha() != null ? bo.getHybridAlpha() : 
+                              (knowledgeInfoVo.getHybridAlpha() != null ? knowledgeInfoVo.getHybridAlpha() : 0.5);
+                
+                allResults = calculateRRF(vectorResults, keywordResults, alpha);
+            } catch (Exception e) {
+                log.error("混合检索执行或合并失败，已自动降级回退到纯向量检索", e);
+                allResults = vectorStoreService.search(queryVectorBo);
+            }
+        } else {
+            allResults = vectorStoreService.search(queryVectorBo);
+        }
 
         // 初始化原始排名
         for (int i = 0; i < allResults.size(); i++) {
@@ -230,4 +271,51 @@ public class KnowledgeFragmentServiceImpl implements IKnowledgeFragmentService {
                 .filter(res -> res.getScore() >= threshold)
                 .collect(Collectors.toList());
     }
+
+    /**
+     * RRF (Reciprocal Rank Fusion) 融合算法
+     * 公式: Score = (1-alpha) * (1 / (k + rank_vector)) + alpha * (1 / (k + rank_keyword))
+     */
+    private List<KnowledgeRetrievalVo> calculateRRF(List<KnowledgeRetrievalVo> vectorList, List<KnowledgeRetrievalVo> keywordList, double alpha) {
+        Map<String, KnowledgeRetrievalVo> allMap = new HashMap<>();
+        Map<String, Double> vectorScores = new HashMap<>();
+        Map<String, Double> keywordScores = new HashMap<>();
+
+        int k = 60; // 常用 RRF 常数
+
+        for (int i = 0; i < vectorList.size(); i++) {
+            KnowledgeRetrievalVo vo = vectorList.get(i);
+            allMap.put(vo.getId(), vo);
+            vectorScores.put(vo.getId(), 1.0 / (k + i + 1));
+        }
+
+        for (int i = 0; i < keywordList.size(); i++) {
+            KnowledgeRetrievalVo vo = keywordList.get(i);
+            if (!allMap.containsKey(vo.getId())) {
+                allMap.put(vo.getId(), vo);
+            }
+            keywordScores.put(vo.getId(), 1.0 / (k + i + 1));
+        }
+
+        // 重新计算得分
+        List<KnowledgeRetrievalVo> fusedResults = new ArrayList<>();
+        for (Map.Entry<String, KnowledgeRetrievalVo> entry : allMap.entrySet()) {
+            String id = entry.getKey();
+            double vScore = vectorScores.getOrDefault(id, 0.0);
+            double kScore = keywordScores.getOrDefault(id, 0.0);
+            
+            // 混合分值
+            double finalScore = (1 - alpha) * vScore + alpha * kScore;
+            
+            // 分值归一化/缩放：将 RRF 分值放大到 0-1 范围
+            // 理论单路最大得分为 1/61 ≈ 0.016，乘以 60 使其处于相似度常用区间
+            KnowledgeRetrievalVo vo = entry.getValue();
+            vo.setScore(finalScore * 60.0); 
+            fusedResults.add(vo);
+        }
+
+        // 按融合分数从高到低排序
+        fusedResults.sort((a, b) -> b.getScore().compareTo(a.getScore()));
+        return fusedResults;
+    }
 }