A POSSIBILITY-LOGIC VIEW OF PRESUPPOSITION TRIGGERS IN BREAKING-NEWS TICKERS

This paper develops an interpretable, real-time method for screening conventional inferences in short, fast-moving headline streams. We model background knowledge as a graded context π\piπ and evaluate the backgrounded content of conventional items with two complementary tests: feasibility and robustness (N). Source reports are discounted by reliability and combined with ordered weighted averaging; editorial caution is represented by hedge operators that reshape π\piπ. Trigger behaviour is encoded by functors for four core families iteratives (again), continuity items (still/yet/already), additives (too/also), and factive predicates tied to timelines and focus alternatives. Licensed assertions update the context via min-narrowing with optional normalization. We analyse algorithmic efficiency using piecewise-linear representations and prove safety properties (idempotence, monotonicity, crisp reduction). A small, realistic case study (20 items) illustrates end-to-end computation of and N, calibration diagnostics, and policy trade-offs: hedging predictably trades recall for safety, while threshold scaling yields a tuneable acceptance frontier. The approach supports concise, faithful rationales by exposing the exact overlap regions and dominant sources that drive decisions. We discuss limitations (threshold sensitivity, approximation scope, cross-lingual variation) and outline extensions to multilingual streams and multimodal cues. Overall, the framework offers a compact, auditable gate for fast editorial workflows that improves caution without sacrificing operational speed.