Tagging (#4426)

### What problem does this PR solve?

#4367

### Type of change

- [x] New Feature (non-breaking change which adds functionality)

api/apps/sdk/dataset.py

@@ -73,7 +73,8 @@ def create(tenant_id):
             chunk_method:
               type: string
               enum: ["naive", "manual", "qa", "table", "paper", "book", "laws",
-                     "presentation", "picture", "one", "knowledge_graph", "email"]
+                     "presentation", "picture", "one", "knowledge_graph", "email", "tag"
+                     ]
               description: Chunking method.
             parser_config:
               type: object
@@ -108,6 +109,7 @@ def create(tenant_id):
         "one",
         "knowledge_graph",
         "email",
+        "tag"
     ]
     check_validation = valid(
         permission,
@@ -302,7 +304,8 @@ def update(tenant_id, dataset_id):
             chunk_method:
               type: string
               enum: ["naive", "manual", "qa", "table", "paper", "book", "laws",
-                     "presentation", "picture", "one", "knowledge_graph", "email"]
+                     "presentation", "picture", "one", "knowledge_graph", "email", "tag"
+                     ]
               description: Updated chunking method.
             parser_config:
               type: object
@@ -339,6 +342,7 @@ def update(tenant_id, dataset_id):
         "one",
         "knowledge_graph",
         "email",
+        "tag"
     ]
     check_validation = valid(
         permission,

api/apps/sdk/dify_retrieval.py

@@ -16,6 +16,7 @@
 from flask import request, jsonify
 
 from api.db import LLMType, ParserType
+from api.db.services.dialog_service import label_question
 from api.db.services.knowledgebase_service import KnowledgebaseService
 from api.db.services.llm_service import LLMBundle
 from api import settings
@@ -54,7 +55,8 @@ def retrieval(tenant_id):
             page_size=top,
             similarity_threshold=similarity_threshold,
             vector_similarity_weight=0.3,
-            top=top
+            top=top,
+            rank_feature=label_question(question, [kb])
         )
         records = []
         for c in ranks["chunks"]:

api/apps/sdk/doc.py

@@ -16,7 +16,7 @@
 import pathlib
 import datetime
 
-from api.db.services.dialog_service import keyword_extraction
+from api.db.services.dialog_service import keyword_extraction, label_question
 from rag.app.qa import rmPrefix, beAdoc
 from rag.nlp import rag_tokenizer
 from api.db import LLMType, ParserType
@@ -276,6 +276,7 @@ def update_doc(tenant_id, dataset_id, document_id):
             "one",
             "knowledge_graph",
             "email",
+            "tag"
         }
         if req.get("chunk_method") not in valid_chunk_method:
             return get_error_data_result(
@@ -1355,6 +1356,7 @@ def retrieval_test(tenant_id):
             doc_ids,
             rerank_mdl=rerank_mdl,
             highlight=highlight,
+            rank_feature=label_question(question, kbs)
         )
         for c in ranks["chunks"]:
             c.pop("vector", None)