Knowledge

The KnowledgeResource handles querying, ingestion, relationship listing, and graph visualization.

Access


user = client.for_user("user-123")
knowledge = user.knowledge

Methods

`query(query, **kwargs)`

Query the knowledge graph with semantic search, optional reranking, and graph reasoning.


results = knowledge.query("What does the user prefer?")
print(results.context)  # Assembled context for LLM prompts
 
for r in results:
    print(f"{r.content} (score: {r.relevance_score})")

Parameter	Type	Default	Description
`query`	`str`	—	Natural language search query
`top_k`	`int`	`10`	Maximum number of results
`use_cache`	`bool`	`True`	Enable semantic cache
`use_rerank`	`bool`	`True`	Enable cross-encoder reranking
`use_emotional_ranking`	`bool`	`True`	Apply emotional scoring
`use_graph_reasoning`	`bool`	`False`	Enable multi-hop graph traversal
`graph_max_hops`	`int`	`3`	Maximum traversal depth
`graph_beam_width`	`int`	`5`	Beam search width for graph traversal
`include_community_context`	`bool`	`False`	Enrich results with community data
`filter_by_community_id`	`str \| None`	`None`	Filter by specific community
`complexity_override`	`str \| None`	`None`	Override complexity: `"simple"`, `"medium"`, or `"complex"`

Returns: QueryResponse with fields:

results — List of QueryResult objects
context — Assembled context string for LLM prompts
query_complexity — Detected complexity ("simple", "medium", "complex")
cache_hit — Whether the result was served from cache
result_count — Total number of results
reasoning_chains — Multi-hop reasoning paths (when graph reasoning enabled)
citations — Source entity citations
timing — TimingBreakdown with per-stage latencies

`ingest(documents)`

Ingest content into the knowledge graph. Entity and relationship extraction is automatic.


# Single string
knowledge.ingest("Alice is a software engineer at Acme Corp")
 
# Multiple documents
knowledge.ingest([
    "User prefers dark mode",
    "Meeting scheduled for Friday",
])
 
# Document objects with metadata
from memorer import Document
knowledge.ingest(Document(
    content="Quarterly report...",
    metadata={"source": "report"},
))

Parameter	Type	Description
`documents`	`str \| dict \| Document \| list`	Content to ingest. Accepts a string, dict with `content` key, `Document` object, or a list of any of these

Returns: IngestResponse with fields entities_created, relationships_created, episodes_created, processing_time_ms, status

`relationships(limit, offset)`

List relationships in the knowledge graph.


rels = knowledge.relationships(limit=50)
for rel in rels.relationships:
    print(f"{rel.source_id} --{rel.relationship_type}--> {rel.target_id}")

Parameter	Type	Default	Description
`limit`	`int`	`100`	Maximum number to return
`offset`	`int`	`0`	Pagination offset

Returns: RelationshipList with a relationships list of Relationship objects

`graph(limit, include_embeddings)`

Get full graph data for visualization. Compatible with D3.js, vis.js, Cytoscape, etc.


viz = knowledge.graph(limit=200)
print(f"Nodes: {len(viz.nodes)}")
print(f"Edges: {len(viz.edges)}")
print(f"Communities: {len(viz.communities)}")
 
if viz.stats:
    print(f"Total entities: {viz.stats.total_entities}")

Parameter	Type	Default	Description
`limit`	`int`	`500`	Maximum number of nodes
`include_embeddings`	`bool`	`False`	Include embedding vectors

Returns: GraphVisualization with fields:

nodes — List of GraphNode (id, label, type, importance, community_id, metadata)
edges — List of GraphEdge (source, target, relationship, weight)
communities — List of GraphCommunity (id, label, summary, entity_count, entity_ids)
stats — Optional KnowledgeStats