Skip to main content

AI Matching

The most significant challenge in digital intelligence is avoiding False Positives. A common name like "John Smith" can trigger thousands of matches. WeCheck solves this by using a multi-signal AI Matching Engine.

Visual vs. Behavioral Signals​

WeCheck doesn't just match names; we match individualities using two primary layers:

1. Visual Anchoring (Facial Recognition)​

If a reference image is provided, our engine uses deep-learning facial comparison models to verify that the profile found actually belongs to the subject. This visual anchor is the strongest signal for identity verification.

2. Behavioral Linkage​

If a photo isn't available, we look for "Behavioral Fingerprints":

  • Handle Consistency: Does the subject use the same or derived handles across different platforms?
  • Biographic Alignment: Do the location mentions, interests, and professional history mentioned in one profile align with the others?
  • Network Proximity: Does the subject interact with the same group of people or entities across different digital networks?

The Confidence Score​

Every match in a WeCheck report is accompanied by a Confidence Score.

  • High Confidence (80-100%): Strong visual or multiple behavioral overlaps.
  • Medium Confidence (50-79%): Narrative overlaps but lacks a visual anchor.
  • Low Confidence (< 50%): Name-matches only; suggested as a "Potential" match for manual verification.

Preventing Bias​

WeCheck's AI is regularly audited for algorithmic bias to ensure that risk flagging is based on Behavioral Sentiment and Verifiable Actions, rather than demographic attributes.