15-20 sources per investigation
Concepts

Trust Score

By Seme Research Team · Updated May 22, 2026

Definition

A Trust Score is a confidence percentage assigned to an investigation dossier based on the quality, quantity, and corroboration of its sources. The score ranges from 0% (no verified information) to 100% (all facts fully corroborated by multiple high-quality sources). The calculation considers: (1) Source Quality — weighted by evidence classification tier (E1=5x, E2=3x, E3=2x, E4=1x, E5=0.5x); (2) Corroboration Rate — percentage of facts supported by 2+ independent sources; (3) Coverage Completeness — how many dossier sections contain verified data; (4) Consistency — absence of conflicting facts. Scores above 80% indicate high confidence suitable for decision-making; 60-80% indicates moderate confidence with some gaps; below 60% indicates significant verification gaps.

How It Works

The trust score algorithm processes all dossier facts through a weighted scoring formula. Each fact receives a base score based on its evidence tier (E1=10 points, E2=6 points, E3=4 points, E4=2 points, E5=1 point). Facts corroborated by multiple independent sources receive a multiplier (2 sources=1.5x, 3+ sources=2x). The total fact score is divided by the maximum possible score (if all facts were E1 with 3+ corroboration) to produce the percentage. The algorithm also penalizes contradictions: each conflicting fact pair reduces the score by 3-5 points. Missing data categories reduce coverage completeness, which caps the maximum achievable score.

Example

A dossier contains 25 facts: 8 E1 facts (all corroborated), 10 E2 facts (7 corroborated), 5 E3 facts (3 corroborated), and 2 E4 facts (none corroborated). The weighted score: (8×10×2 + 7×6×1.5 + 3×4×1.5 + 2×2×1) = 160 + 63 + 18 + 4 = 245. Maximum possible: 25×10×2 = 500. Raw score: 245/500 = 49%. Coverage bonus for 6/8 sections populated: +15%. One contradiction penalty: -4%. Final trust score: 60% — moderate confidence with significant gaps in E3-E5 verification.

Applications

  • Automated quality assessment for investigation reports
  • Risk scoring for investment and lending decisions
  • Compliance threshold verification for regulatory requirements
  • Comparative analysis of multiple investigation targets

Key Statistics

MetricValueSource
Score Range0-100%Seme methodology
High Confidence Threshold≥80%Seme methodology
E1 Source Weight5xScoring algorithm
Average Score (Production)78%Seme analytics

Related Terms

Platform Data

15-20
Sources/Investigation
78%
Avg Trust Score
25+
Glossary Terms
10-30 min
Investigation Time

Related Resources