Burst Detection

Kleinberg burst detection — terms with a sudden surge in usage.

Burst Detection

The Burst Detection widget.

Overview

Applies Kleinberg’s burst-detection algorithm to keyword (or other entity) frequencies over time to find terms that experienced a sharp, sustained surge — the markers of emerging research fronts. Returns each burst’s start/end year and intensity (weight); selecting a burst outputs its documents.

Inputs

  • Data (Table) — keywords + Year.

Outputs

  • Bursts (Table) — Keyword, Start, End, Weight.

  • Selected Documents (Table) — documents of the selected bursting term(s).

Controls

  • Keywords column — the entity field to scan for bursts.

  • Top N — limit to the N strongest bursts.

  • s (burst scaling) — the state-transition intensity factor (higher = only stronger bursts).

  • gamma (transition cost) — cost of switching into a burst state (higher = fewer, longer bursts).

  • Min duration (yrs) — discard bursts shorter than this.

Actions: Detect Bursts.