AI/dify
1
0
Fork 0
mirror of https://git.mirrors.martin98.com/https://github.com/langgenius/dify.git synced 2025-08-12 16:59:04 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix weaviate hybrid search issue (#1600)

Browse Source 
Co-authored-by: jyong <jyong@dify.ai>
This commit is contained in:
Jyong 2023-11-22 16:41:20 +08:00 committed by GitHub
parent 9587479b76
commit b930716745
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
5 changed files with 11 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
api/core/index/vector_index/weaviate_vector_index.py
View File

@ -111,7 +111,7 @@ class WeaviateVectorIndex(BaseVectorIndex):
if self._vector_store:
return self._vector_store
attributes = ['doc_id', 'dataset_id', 'document_id']
attributes = ['doc_id', 'dataset_id', 'document_id', 'doc_hash']
if self._is_origin():
attributes = ['doc_id']

11
api/core/vector_store/vector/weaviate.py
View File

@ -60,7 +60,7 @@ def _create_weaviate_client(**kwargs: Any) -> Any:
def _default_score_normalizer(val: float) -> float:
return 1 - 1 / (1 + np.exp(val))
return 1 - val
def _json_serializable(value: Any) -> Any:
@ -243,7 +243,8 @@ class Weaviate(VectorStore):
query_obj = query_obj.with_where(kwargs.get("where_filter"))
if kwargs.get("additional"):
query_obj = query_obj.with_additional(kwargs.get("additional"))
result = query_obj.with_bm25(query=content).with_limit(k).do()
properties = ['text', 'dataset_id', 'doc_hash', 'doc_id', 'document_id']
result = query_obj.with_bm25(query=query, properties=properties).with_limit(k).do()
if "errors" in result:
raise ValueError(f"Error during query: {result['errors']}")
docs = []
@ -380,14 +381,14 @@ class Weaviate(VectorStore):
result = (
query_obj.with_near_vector(vector)
.with_limit(k)
.with_additional("vector")
.with_additional(["vector", "distance"])
.do()
)
else:
result = (
query_obj.with_near_text(content)
.with_limit(k)
.with_additional("vector")
.with_additional(["vector", "distance"])
.do()
)
@ -397,7 +398,7 @@ class Weaviate(VectorStore):
docs_and_scores = []
for res in result["data"]["Get"][self._index_name]:
text = res.pop(self._text_key)
score = np.dot(res["_additional"]["vector"], embedded_query)
score = res["_additional"]["distance"]
docs_and_scores.append((Document(page_content=text, metadata=res), score))
return docs_and_scores

2
api/core/vector_store/weaviate_vector_store.py
View File

@ -1,4 +1,4 @@
from langchain.vectorstores import Weaviate
from core.vector_store.vector.weaviate import Weaviate
class WeaviateVectorStore(Weaviate):

4
docker/docker-compose.middleware.yaml
View File

@ -30,7 +30,7 @@ services:
# The Weaviate vector store.
weaviate:
image: semitechnologies/weaviate:1.18.4
image: semitechnologies/weaviate:1.19.0
restart: always
volumes:
# Mount the Weaviate data directory to the container.
@ -63,4 +63,4 @@ services:
# environment:
# QDRANT__API_KEY: 'difyai123456'
# ports:
# - "6333:6333"
# - "6333:6333"

2
docker/docker-compose.yaml
View File

@ -253,7 +253,7 @@ services:
# The Weaviate vector store.
weaviate:
image: semitechnologies/weaviate:1.18.4
image: semitechnologies/weaviate:1.19.0
restart: always
volumes:
# Mount the Weaviate data directory to the container.