Back to App

How Verification Works

Confidence-Based Verification

GroundTruth uses a multi-signal verification system that calculates a confidence score (0-100%) for each reported incident. This score reflects how much corroboration exists for the report, not whether the report is officially confirmed.

Confidence Tiers

Reports are categorized into three tiers based on their confidence score:

80-100%

High Confidence

Multiple corroborating reports. Full audio and vibration alerts. Displayed prominently on map.

60-79%

Medium Confidence

Some corroboration exists. Optional alerts (user preference). Displayed on map with moderate visibility.

30-59%

Low Confidence

Limited corroboration. No alerts triggered. Displayed on map only for awareness.

What Affects Confidence?

Multiple factors contribute to the confidence score calculation:

#

Report Count

More unique reporters corroborating the same incident increases confidence.

T

Time Clustering

Reports submitted close together in time (within 15 minutes) add a clustering bonus.

S

Spatial Clustering

Reports from nearby locations (tight clustering under 200m) increase confidence more than spread-out reports.

C

Clear Reports

Users can submit "clear" reports if they pass a location and don't observe the reported condition. These reduce confidence.

D

Time Decay

Confidence naturally decreases over time. Lower-confidence reports decay faster than high-confidence ones.

Why Alerts Differ

You may notice that some reports trigger alerts while others don't. This is intentional:

What Verification Means (and Doesn't Mean)

Verification DOES mean:


Verification DOES NOT mean:

Note: All information is provided for situational awareness only. User discretion is advised. See our Terms and Disclaimer for more information.

Sensitive Areas

Some locations are designated as sensitive zones (near schools, hospitals, shelters). Reports in these areas require higher confidence thresholds before triggering alerts, adding an extra layer of verification to reduce potential false positives in sensitive contexts.

How You Can Help

The system works best when community members participate responsibly:

Back to Home