Struct ImplicitConflictDetector 

pub struct ImplicitConflictDetector { /* private fields */ }

Write-time implicit conflict detector.

Compares a new edge’s predicate against existing active edges on the same source entity using the configured similarity method and threshold.

Examples

use zeph_config::ImplicitConflictConfig;
use zeph_memory::graph::implicit_conflict::ImplicitConflictDetector;

let config = ImplicitConflictConfig { enabled: true, ..Default::default() };
let detector = ImplicitConflictDetector::new(config);
let candidates = detector.detect_candidates(42, "employ", &[(1, "employs")], false);
assert!(!candidates.is_empty());

Implementations

impl ImplicitConflictDetector

pub fn new(config: ImplicitConflictConfig) -> Self

Create a new detector with the given configuration.

pub fn detect_candidates( &self, new_edge_id: i64, new_predicate: &str, existing: &[(i64, &str)], is_cardinality_n: bool, ) -> Vec<ConflictCandidate>

Detect implicit conflict candidates for a new predicate against existing ones.

Returns pairs where normalized Levenshtein similarity is >= conflict_similarity_threshold and the predicates differ (identical predicates are already handled by APEX-MEM explicit supersession).

Returns an empty vec when enabled = false or when the cardinality flag is_cardinality_n is set (FR-011).

Arguments

• new_edge_id — database ID of the newly inserted edge
• new_predicate — canonical relation of the new edge
• existing — slice of (edge_id, canonical_relation) for all other active edges on the same source entity
• is_cardinality_n — set to true for multi-valued predicates; skips detection

pub async fn stage_candidates( &self, candidates: &[ConflictCandidate], tx: &mut DbTransaction<'_>, ttl_days: u32, ) -> Result<(), MemoryError>

Persist conflict candidates into implicit_conflict_candidates.

Each candidate is inserted with status = 'pending' and an expiry of now + ttl_days * 86400 seconds.

Errors

Returns a MemoryError on database write failure.

pub fn normalized_levenshtein(a: &str, b: &str) -> f64

Compute normalized Levenshtein similarity between two strings.

Returns a value in [0.0, 1.0] where 1.0 means identical. Returns 1.0 if both strings are empty, 0.0 if only one is empty.

pub fn is_enabled(&self) -> bool

Returns true when detection is enabled.

pub fn candidate_ttl_days(&self) -> u32

Returns the configured TTL for conflict candidates, in days.