The Veracity engine measures how faithfully an AI model represents cited web sources by calculating a mathematical ratio of survived facts over the sum of survived, lost, and distorted claims, persisting individual and aggregate scores to dedicated database tables.
flowchart LR A[Grounded Search Response] --> B[Extract Cited URLs] B --> C[Fetch Source Page Content] C --> D[Evaluate Claim Fidelity] D --> E[Survived Claims] D --> F[Lost Claims] D --> G[Distorted Claims] E --> H[Fidelity Score Calculation] F --> H G --> H H --> I[(Database Tables)]
