diff --git a/backend/Dockerfile b/backend/Dockerfile
index c36f8ce2e..4488129d9 100644
--- a/backend/Dockerfile
+++ b/backend/Dockerfile
@@ -16,9 +16,10 @@ RUN apt-get update && \
# Set LD_LIBRARY_PATH
ENV LD_LIBRARY_PATH=/usr/lib/x86_64-linux-gnu:$LD_LIBRARY_PATH
# Copy requirements file and install Python dependencies
-COPY requirements.txt /code/
+COPY requirements.txt constraints.txt /code/
# --no-cache-dir --upgrade
-RUN pip install -r requirements.txt
+RUN pip install --upgrade pip
+RUN pip install -r requirements.txt -c constraints.txt
# Copy application code
COPY . /code
# Set command
diff --git a/backend/constraints.txt b/backend/constraints.txt
new file mode 100644
index 000000000..2c785f6de
--- /dev/null
+++ b/backend/constraints.txt
@@ -0,0 +1,4 @@
+-f https://download.pytorch.org/whl/torch_stable.html
+torch==2.3.1+cpu
+torchvision==0.18.1+cpu
+torchaudio==2.3.1+cpu
\ No newline at end of file
diff --git a/backend/score.py b/backend/score.py
index 2318c22e8..e2a6c4e51 100644
--- a/backend/score.py
+++ b/backend/score.py
@@ -31,11 +31,10 @@
from src.ragas_eval import *
from starlette.types import ASGIApp, Receive, Scope, Send
from langchain_neo4j import Neo4jGraph
-from src.entities.source_node import sourceNode
from starlette.middleware.sessions import SessionMiddleware
-from starlette.responses import HTMLResponse, RedirectResponse,JSONResponse
from starlette.requests import Request
-import secrets
+from dotenv import load_dotenv
+load_dotenv(override=True)
logger = CustomLogger()
CHUNK_DIR = os.path.join(os.path.dirname(__file__), "chunks")
diff --git a/backend/src/llm.py b/backend/src/llm.py
index b83773f42..ef835b560 100644
--- a/backend/src/llm.py
+++ b/backend/src/llm.py
@@ -180,13 +180,16 @@ async def get_graph_document_list(
else:
node_properties = ["description"]
relationship_properties = ["description"]
+ TOOL_SUPPORTED_MODELS = {"qwen3", "deepseek"}
+ model_name = llm.model_name.lower()
+ ignore_tool_usage = not any(pattern in model_name for pattern in TOOL_SUPPORTED_MODELS)
llm_transformer = LLMGraphTransformer(
llm=llm,
node_properties=node_properties,
relationship_properties=relationship_properties,
allowed_nodes=allowedNodes,
allowed_relationships=allowedRelationship,
- ignore_tool_usage=True,
+ ignore_tool_usage=ignore_tool_usage,
additional_instructions=ADDITIONAL_INSTRUCTIONS+ (additional_instructions if additional_instructions else "")
)
diff --git a/backend/src/main.py b/backend/src/main.py
index c2f2a80f3..4bdb6ba51 100644
--- a/backend/src/main.py
+++ b/backend/src/main.py
@@ -665,39 +665,40 @@ def upload_file(graph, model, chunk, chunk_number:int, total_chunks:int, origina
return f"Chunk {chunk_number}/{total_chunks} saved"
def get_labels_and_relationtypes(uri, userName, password, database):
- excluded_labels = {'Document', 'Chunk', '_Bloom_Perspective_', '__Community__', '__Entity__', 'Session', 'Message'}
- excluded_relationships = {
- 'PART_OF', 'NEXT_CHUNK', 'HAS_ENTITY', '_Bloom_Perspective_', 'FIRST_CHUNK',
- 'SIMILAR', 'IN_COMMUNITY', 'PARENT_COMMUNITY', 'NEXT', 'LAST_MESSAGE'}
- driver = get_graphDB_driver(uri, userName, password,database)
- with driver.session(database=database) as session:
- result = session.run("CALL db.schema.visualization() YIELD nodes, relationships RETURN nodes, relationships")
- if not result:
- return []
- record = result.single()
- nodes = record["nodes"]
- relationships = record["relationships"]
- node_map = {}
- for node in nodes:
- node_id = node.element_id
- labels = list(node.labels)
- if labels:
- node_map[node_id] = ":".join(labels)
- triples = []
- for rel in relationships:
- start_id = rel.start_node.element_id
- end_id = rel.end_node.element_id
- rel_type = rel.type
- start_label = node_map.get(start_id)
- end_label = node_map.get(end_id)
- if start_label and end_label:
- if (
- start_label not in excluded_labels and
- end_label not in excluded_labels and
- rel_type not in excluded_relationships
- ):
- triples.append(f"{start_label}-{rel_type}->{end_label}")
- return {"triplets" : list(set(triples))}
+ excluded_labels = {'Document', 'Chunk', '_Bloom_Perspective_', '__Community__', '__Entity__', 'Session', 'Message'}
+ excluded_relationships = {
+ 'NEXT_CHUNK', '_Bloom_Perspective_', 'FIRST_CHUNK',
+ 'SIMILAR', 'IN_COMMUNITY', 'PARENT_COMMUNITY', 'NEXT', 'LAST_MESSAGE'
+ }
+ driver = get_graphDB_driver(uri, userName, password,database)
+ triples = set()
+ with driver.session(database=database) as session:
+ result = session.run("""
+ MATCH (n)-[r]->(m)
+ RETURN DISTINCT labels(n) AS fromLabels, type(r) AS relType, labels(m) AS toLabels
+ """)
+ for record in result:
+ from_labels = record["fromLabels"]
+ to_labels = record["toLabels"]
+ rel_type = record["relType"]
+ from_label = next((lbl for lbl in from_labels if lbl not in excluded_labels), None)
+ to_label = next((lbl for lbl in to_labels if lbl not in excluded_labels), None)
+ if not from_label or not to_label:
+ continue
+ if rel_type == 'PART_OF':
+ if from_label == 'Chunk' and to_label == 'Document':
+ continue
+ elif rel_type == 'HAS_ENTITY':
+ if from_label == 'Chunk':
+ continue
+ elif (
+ from_label in excluded_labels or
+ to_label in excluded_labels or
+ rel_type in excluded_relationships
+ ):
+ continue
+ triples.add(f"{from_label}-{rel_type}->{to_label}")
+ return {"triplets": list(triples)}
def manually_cancelled_job(graph, filenames, source_types, merged_dir, uri):
diff --git a/backend/src/make_relationships.py b/backend/src/make_relationships.py
index a07f29c9c..97aa7e33e 100644
--- a/backend/src/make_relationships.py
+++ b/backend/src/make_relationships.py
@@ -1,6 +1,7 @@
from langchain_neo4j import Neo4jGraph
from langchain.docstore.document import Document
from src.shared.common_fn import load_embedding_model,execute_graph_query
+from src.shared.common_fn import load_embedding_model,execute_graph_query
import logging
from typing import List
import os
@@ -34,6 +35,7 @@ def merge_relationship_between_chunk_and_entites(graph: Neo4jGraph, graph_docume
MERGE (c)-[:HAS_ENTITY]->(n)
"""
execute_graph_query(graph,unwind_query, params={"batch_data": batch_data})
+ execute_graph_query(graph,unwind_query, params={"batch_data": batch_data})
def create_chunk_embeddings(graph, chunkId_chunkDoc_list, file_name):
@@ -60,6 +62,7 @@ def create_chunk_embeddings(graph, chunkId_chunkDoc_list, file_name):
MERGE (c)-[:PART_OF]->(d)
"""
execute_graph_query(graph,query_to_create_embedding, params={"fileName":file_name, "data":data_for_query})
+ execute_graph_query(graph,query_to_create_embedding, params={"fileName":file_name, "data":data_for_query})
def create_relation_between_chunks(graph, file_name, chunks: List[Document])->list:
logging.info("creating FIRST_CHUNK and NEXT_CHUNK relationships between chunks")
@@ -128,6 +131,7 @@ def create_relation_between_chunks(graph, file_name, chunks: List[Document])->li
MERGE (c)-[:PART_OF]->(d)
"""
execute_graph_query(graph,query_to_create_chunk_and_PART_OF_relation, params={"batch_data": batch_data})
+ execute_graph_query(graph,query_to_create_chunk_and_PART_OF_relation, params={"batch_data": batch_data})
query_to_create_FIRST_relation = """
UNWIND $relationships AS relationship
@@ -137,6 +141,7 @@ def create_relation_between_chunks(graph, file_name, chunks: List[Document])->li
MERGE (d)-[:FIRST_CHUNK]->(c))
"""
execute_graph_query(graph,query_to_create_FIRST_relation, params={"f_name": file_name, "relationships": relationships})
+ execute_graph_query(graph,query_to_create_FIRST_relation, params={"f_name": file_name, "relationships": relationships})
query_to_create_NEXT_CHUNK_relation = """
UNWIND $relationships AS relationship
@@ -153,7 +158,7 @@ def create_relation_between_chunks(graph, file_name, chunks: List[Document])->li
def create_chunk_vector_index(graph):
start_time = time.time()
try:
- vector_index_query = "SHOW INDEXES YIELD * WHERE labelsOrTypes = ['Chunk'] and type = 'VECTOR' AND name = 'vector' return options"
+ vector_index_query = "SHOW INDEXES YIELD name, type, labelsOrTypes, properties WHERE name = 'vector' AND type = 'VECTOR' AND 'Chunk' IN labelsOrTypes AND 'embedding' IN properties RETURN name"
vector_index = execute_graph_query(graph,vector_index_query)
if not vector_index:
vector_store = Neo4jVector(embedding=EMBEDDING_FUNCTION,
@@ -168,7 +173,7 @@ def create_chunk_vector_index(graph):
else:
logging.info(f"Index already exist,Skipping creation. Time taken: {time.time() - start_time:.2f} seconds")
except Exception as e:
- if "EquivalentSchemaRuleAlreadyExists" in str(e):
+ if ("EquivalentSchemaRuleAlreadyExists" in str(e) or "An equivalent index already exists" in str(e)):
logging.info("Vector index already exists, skipping creation.")
else:
raise
\ No newline at end of file
diff --git a/frontend/index.html b/frontend/index.html
index 85618e155..c669468ee 100644
--- a/frontend/index.html
+++ b/frontend/index.html
@@ -1,16 +1,20 @@
-
-
-
-
-
-
-
- Neo4j graph builder
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+ Neo4j graph builder
+
+
+
+
+
+
+
+
+