refactor: ♻️ 对RAG的ETL流程进行再次优化

Author: StephenQiu30

.gitattributes

@@ -87,3 +87,6 @@ nacos-config/common-secret-prod.properties
 .env
 .env.*
 !.env.example
+
+# Data
+data/

.gitignore

@@ -80,4 +80,10 @@ public class AiChatRecordDTO implements Serializable {
      */
     @Schema(description = "生成消耗 token")
     private Integer completionTokens;
+
+    /**
+     * 检索元数据（来源分片、分数等）
+     */
+    @Schema(description = "检索元数据")
+    private String retrievalMetadata;
 }

algorithm-api/algorithm-api-ai/src/main/java/com/stephen/cloud/api/ai/model/dto/AiChatRecordDTO.java

@@ -38,5 +38,17 @@ public class DocumentChunkQueryRequest extends PageRequest implements Serializab
      */
     @Schema(description = "分片序号")
     private Integer chunkIndex;
+
+    /**
+     * 标签过滤 (支持模糊匹配)
+     */
+    @Schema(description = "标签过滤")
+    private String tags;
+
+    /**
+     * 是否包含代码
+     */
+    @Schema(description = "是否包含代码")
+    private Boolean hasCode;
 }
 

algorithm-api/algorithm-api-ai/src/main/java/com/stephen/cloud/api/knowledge/model/dto/chunk/DocumentChunkAddRequest.java

@@ -29,5 +29,29 @@ public class KnowledgeBaseAddRequest implements Serializable {
      */
     @Schema(description = "知识库描述")
     private String description;
+
+    /**
+     * 向量相似度模式 (cosine/euclidean/dot_product)
+     */
+    @Schema(description = "向量相似度模式")
+    private String similarityMode;
+
+    /**
+     * 分片策略 (token/recursive/semantic)
+     */
+    @Schema(description = "分片策略")
+    private String chunkStrategy;
+
+    /**
+     * 分片大小
+     */
+    @Schema(description = "分片大小")
+    private Integer chunkSize;
+
+    /**
+     * 分片重叠
+     */
+    @Schema(description = "分片重叠")
+    private Integer chunkOverlap;
 }
 

algorithm-api/algorithm-api-ai/src/main/java/com/stephen/cloud/api/knowledge/model/dto/chunk/DocumentChunkQueryRequest.java

@@ -44,5 +44,29 @@ public class KnowledgeBaseUpdateRequest implements Serializable {
      */
     @Schema(description = "状态 (0-启用, 1-禁用等)")
     private Integer status;
+
+    /**
+     * 向量相似度模式
+     */
+    @Schema(description = "向量相似度模式")
+    private String similarityMode;
+
+    /**
+     * 分片策略
+     */
+    @Schema(description = "分片策略")
+    private String chunkStrategy;
+
+    /**
+     * 分片大小
+     */
+    @Schema(description = "分片大小")
+    private Integer chunkSize;
+
+    /**
+     * 分片重叠
+     */
+    @Schema(description = "分片重叠")
+    private Integer chunkOverlap;
 }
 

algorithm-api/algorithm-api-ai/src/main/java/com/stephen/cloud/api/knowledge/model/dto/chunk/DocumentChunkUpdateRequest.java

@@ -45,4 +45,16 @@ public class KnowledgeDocumentAddRequest implements Serializable {
      */
     @Schema(description = "文件大小 (字节)")
     private Long sizeBytes;
+
+    /**
+     * 文档标签 (逗号分隔，如：算法,动态规划,LeetCode)
+     */
+    @Schema(description = "文档标签")
+    private String tags;
+
+    /**
+     * 是否包含代码
+     */
+    @Schema(description = "是否包含代码")
+    private Boolean hasCode;
 }

algorithm-api/algorithm-api-ai/src/main/java/com/stephen/cloud/api/knowledge/model/dto/knowledgebase/KnowledgeBaseAddRequest.java

@@ -45,4 +45,16 @@ public class KnowledgeDocumentUpdateRequest implements Serializable {
      */
     @Schema(description = "错误信息")
     private String errorMsg;
+
+    /**
+     * 文档标签 (逗号分隔)
+     */
+    @Schema(description = "文档标签")
+    private String tags;
+
+    /**
+     * 是否包含代码
+     */
+    @Schema(description = "是否包含代码")
+    private Boolean hasCode;
 }