Algorithm 1. Stateful Complex Event Matching in GICEDCAM.

Input: stream of spatial events e = ⟨type, bindings, t, conf, corrected⟩    window W, watermark ω State: partial matches M[q] keyed by (q, bindings) upon spatial event e at time t:  // events are pre-filtered & projected  // 1) advance existing partials  for each (q, b) in M where δ(q, e.type, b, t) is enabled    q′ ← δ(q, e.type, b, t)  // checks Allen relation & gap constraints    b′ ← unify(b, e.bindings)  // variable binding; reject on conflict    upsert M[q′] with {bindings = b′, start_time(b), last_time = t}  // 2) possibly start a new partial  if δ(q0, e.type, e.bindings, t) enabled    upsert M[q1] with {bindings = e.bindings, start_time = t, last_time = t}  // 3) emit completes when safe  for each partial in M with state q ∈ F    if (last_time − start_time) ≤ W and last_time ≤ ω      emit ComplexEvent(partial.bindings, [start_time, last_time])      delete partial  // 4) prune stale partials (semantic bound)  delete any partial with (current_time − start_time) > W