gardel/lightrag
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Simplified graph retrival by removing inclusive and min_degree parameters

Browse Source
This commit is contained in:
yangdx
2025-04-02 16:27:58 +08:00
parent df07c2a8b1
commit 4994bd1461
2 changed files with 16 additions and 36 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
lightrag/api/routers/graph_routes.py
View File

@@ -41,7 +41,7 @@ def create_graph_routes(rag, api_key: Optional[str] = None):
label (str): Label to get knowledge graph for label (str): Label to get knowledge graph for
max_depth (int, optional): Maximum depth of graph. Defaults to 3. max_depth (int, optional): Maximum depth of graph. Defaults to 3.
inclusive_search (bool, optional): If True, search for nodes that include the label. Defaults to False. inclusive_search (bool, optional): If True, search for nodes that include the label. Defaults to False.
min_degree (int, optional): Minimum degree of nodes. Defaults to 0. min_degree (int, optional): Minimum degree of nodes. Defaults to 0. （Deprecated, always 0）
Returns: Returns:
Dict[str, List[str]]: Knowledge graph for label Dict[str, List[str]]: Knowledge graph for label
@@ -49,8 +49,6 @@ def create_graph_routes(rag, api_key: Optional[str] = None):
return await rag.get_knowledge_graph( return await rag.get_knowledge_graph(
node_label=label, node_label=label,
max_depth=max_depth, max_depth=max_depth,
inclusive=inclusive,
min_degree=min_degree,
) )
return router return router

48
lightrag/kg/neo4j_impl.py
View File

@@ -614,22 +614,19 @@ class Neo4JStorage(BaseGraphStorage):
self, self,
node_label: str, node_label: str,
max_depth: int = 3, max_depth: int = 3,
min_degree: int = 0, max_nodes: int = MAX_GRAPH_NODES,
inclusive: bool = False,
) -> KnowledgeGraph: ) -> KnowledgeGraph:
""" """
Retrieve a connected subgraph of nodes where the label includes the specified `node_label`. Retrieve a connected subgraph of nodes where the label includes the specified `node_label`.
Maximum number of nodes is constrained by the environment variable `MAX_GRAPH_NODES` (default: 1000). Maximum number of nodes is constrained by the environment variable `MAX_GRAPH_NODES` (default: 1000).
When reducing the number of nodes, the prioritization criteria are as follows: When reducing the number of nodes, the prioritization criteria are as follows:
1. min_degree does not affect nodes directly connected to the matching nodes 1. Label matching nodes take precedence
2. Label matching nodes take precedence 2. Followed by nodes directly connected to the matching nodes
3. Followed by nodes directly connected to the matching nodes 3. Finally, the degree of the nodes
4. Finally, the degree of the nodes
Args: Args:
node_label: Label of the starting node node_label: Label of the starting node
max_depth: Maximum depth of the subgraph max_depth: Maximum depth of the subgraph
min_degree: Minimum degree of nodes to include. Defaults to 0
inclusive: Do an inclusive search if true inclusive: Do an inclusive search if true
Returns: Returns:
KnowledgeGraph: Complete connected subgraph for specified node KnowledgeGraph: Complete connected subgraph for specified node
@@ -647,7 +644,6 @@ class Neo4JStorage(BaseGraphStorage):
MATCH (n) MATCH (n)
OPTIONAL MATCH (n)-[r]-() OPTIONAL MATCH (n)-[r]-()
WITH n, COALESCE(count(r), 0) AS degree WITH n, COALESCE(count(r), 0) AS degree
WHERE degree >= $min_degree
ORDER BY degree DESC ORDER BY degree DESC
LIMIT $max_nodes LIMIT $max_nodes
WITH collect({node: n}) AS filtered_nodes WITH collect({node: n}) AS filtered_nodes
@@ -660,18 +656,14 @@ class Neo4JStorage(BaseGraphStorage):
""" """
result_set = await session.run( result_set = await session.run(
main_query, main_query,
{"max_nodes": MAX_GRAPH_NODES, "min_degree": min_degree}, {"max_nodes": max_nodes},
) )
else: else:
# Main query uses partial matching # Main query uses partial matching
main_query = """ main_query = """
MATCH (start) MATCH (start)
WHERE WHERE start.entity_id = $entity_id
CASE
WHEN $inclusive THEN start.entity_id CONTAINS $entity_id
ELSE start.entity_id = $entity_id
END
WITH start WITH start
CALL apoc.path.subgraphAll(start, { CALL apoc.path.subgraphAll(start, {
relationshipFilter: '', relationshipFilter: '',
@@ -684,13 +676,11 @@ class Neo4JStorage(BaseGraphStorage):
UNWIND nodes AS node UNWIND nodes AS node
OPTIONAL MATCH (node)-[r]-() OPTIONAL MATCH (node)-[r]-()
WITH node, COALESCE(count(r), 0) AS degree, start, nodes, relationships WITH node, COALESCE(count(r), 0) AS degree, start, nodes, relationships
WHERE node = start OR EXISTS((start)--(node)) OR degree >= $min_degree
ORDER BY ORDER BY
CASE CASE
WHEN node = start THEN 3 WHEN node = start THEN 0
WHEN EXISTS((start)--(node)) THEN 2 ELSE length(shortestPath((start)--(node)))
ELSE 1 END ASC,
END DESC,
degree DESC degree DESC
LIMIT $max_nodes LIMIT $max_nodes
WITH collect({node: node}) AS filtered_nodes WITH collect({node: node}) AS filtered_nodes
@@ -704,11 +694,9 @@ class Neo4JStorage(BaseGraphStorage):
result_set = await session.run( result_set = await session.run(
main_query, main_query,
{ {
"max_nodes": MAX_GRAPH_NODES, "max_nodes": max_nodes,
"entity_id": node_label, "entity_id": node_label,
"inclusive": inclusive,
"max_depth": max_depth, "max_depth": max_depth,
"min_degree": min_degree,
}, },
) )
@@ -759,19 +747,13 @@ class Neo4JStorage(BaseGraphStorage):
logger.warning( logger.warning(
"Neo4j: falling back to basic Cypher recursive search..." "Neo4j: falling back to basic Cypher recursive search..."
) )
if inclusive:
logger.warning(
"Neo4j: inclusive search mode is not supported in recursive query, using exact matching"
)
return await self._robust_fallback( return await self._robust_fallback(
node_label, max_depth, min_degree node_label, max_depth, max_nodes
) )
return result return result
async def _robust_fallback( async def _robust_fallback(self, node_label: str, max_depth: int, max_nodes: int) -> KnowledgeGraph:
self, node_label: str, max_depth: int, min_degree: int = 0
) -> KnowledgeGraph:
""" """
Fallback implementation when APOC plugin is not available or incompatible. Fallback implementation when APOC plugin is not available or incompatible.
This method implements the same functionality as get_knowledge_graph but uses This method implements the same functionality as get_knowledge_graph but uses
@@ -790,7 +772,7 @@ class Neo4JStorage(BaseGraphStorage):
if current_depth > max_depth: if current_depth > max_depth:
logger.debug(f"Reached max depth: {max_depth}") logger.debug(f"Reached max depth: {max_depth}")
return return
if len(visited_nodes) >= MAX_GRAPH_NODES: if len(visited_nodes) >= max_nodes:
logger.debug(f"Reached max nodes limit: {MAX_GRAPH_NODES}") logger.debug(f"Reached max nodes limit: {MAX_GRAPH_NODES}")
return return
@@ -816,8 +798,8 @@ class Neo4JStorage(BaseGraphStorage):
await results.consume() # Ensure results are consumed await results.consume() # Ensure results are consumed
# Nodes not connected to start node need to check degree # Nodes not connected to start node need to check degree
if current_depth > 1 and len(records) < min_degree: # if current_depth > 1 and len(records) < min_degree:
return # return
# Add current node to result # Add current node to result
result.nodes.append(node) result.nodes.append(node)