A Seven-Layer Model for Standardising AI Fairness Assessment