Inspection Operations¶

Methods for inspecting the internal state of the knowledge graph.

`node_info()`¶

Detailed information about a specific node.

info = r.node_info(label: str) -> NodeInfo

NodeInfo with:

Field	Type	Description
`label`	`str`	Canonical label
`surface_label`	`str`	Display form
`id`	`int`	Unique NodeId
`confidence`	`float`	Confidence score (0.0–1.0)
`tier`	`int`	Tier level (1, 2, or 3)
`status`	`str`	Node status: "New", "Candidate", "Stable", "Deprecated", "Quarantine"
`is_seed`	`bool`	Whether this is an immutable seed atom
`is_locked`	`bool`	Whether the node is protected from deletion
`is_stable`	`bool`	Whether the node has reached Stable status
`compression_state`	`str`	"raw" or "compressed"
`layer`	`int`	Compositional depth
`atoms`	`list[int]`	Atom set as NodeId list
`derived_from_node_ids`	`list[int]`	Source nodes for compressed/derived nodes
`compression_reason`	`str \| None`	Reason for compression

All senses of a concept with grounding evidence and composition details.

senses = r.senses(concept: str) -> list[SenseInfo]

list[SenseInfo] — each entry has:

Field	Type	Description
`sense_idx`	`int`	Sense index within the node
`n_contexts`	`int`	Number of observational contexts
`coherence`	`float`	Internal consistency score
`status`	`str`	"Fragile" or "Mature"
`core_atoms`	`list[str]`	Atom labels for this sense
`layer`	`int`	Compositional depth
`grounding_score`	`float`	Net grounding score
`grounding_evidence`	`GroundingEvidence`	Full evidence trail
`compositions`	`list[tuple[str, int]]`	(label, sense_id) composition pairs
`condition_label`	`str \| None`	Optional annotation

List all known node or atom labels.

all_nodes = r.nodes(include_seeds: bool = False) -> list[str]
all_atoms = r.atoms(include_seeds: bool = False) -> list[str]

Confidence scores for all nodes.

conf_map = r.confidence_map() -> dict[str, float]

Unpromoted tokens with highest centrality scores.

candidates = r.entity_candidates(top_k: int = 10) -> list[tuple[str, float]]

System status including total nodes, atoms, contexts, and configuration parameters.

status = r.status() -> dict[str, float]

Nodes marked for deletion that require explicit approval (high-impact nodes).

pending = r.pending_removals() -> list[int]