- name
- referral-abuse-detection-v1
- description
- Detects referral abuse violations by investigating accounts, analyzing traffic patterns, and scoring violation indicators to determine enforcement action.
- required_tools
- ["investigate_account","analyze_traffic_patterns","determine_enforcement_action"]
- output_fields
- ["enforcement_action"]
### Step 1: Investigate Account
Call `investigate_account` with `account_id`. Record all returned fields: `address_validity`, `email_pattern_suspicious`, `website_verified`, `business_description`, `account_status`, `connected_accounts`, and `login_geographic_consistency`.
### Step 2: Analyze Traffic Patterns
Call `analyze_traffic_patterns` with `account_id`. Record all returned fields: `revenue_amount`, `click_through_rate`, `page_views`, `device_distribution`, `referral_source_quality`, `payment_method_shared`, and `order_patterns_suspicious`.
### Step 3: Get Enforcement Guidelines
Call `determine_enforcement_action` with `account_id` to retrieve enforcement action guidelines.
### Step 4: Calculate Violation Scores and Determine Enforcement Action
Using the data from Steps 1 and 2, calculate four violation scores by counting how many indicators are TRUE for each category:
**Abusive Account Creation Score** (threshold >= 3):
- address_validity is false (+1)
- email_pattern_suspicious is true (+1)
- website_verified is false (+1)
- connected_accounts >= 15 (+1)
- login_geographic_consistency is false (+1)
**Misleading Ad Copy Score** (threshold >= 3):
- website_verified is false (+1)
- referral_source_quality is "Low" or "Medium" (+1)
- order_patterns_suspicious is true (+1)
- click_through_rate > 0.4 (+1)
**Personal Orders Score** (threshold >= 3):
- payment_method_shared is true (+1)
- connected_accounts > 0 AND connected_accounts < 15 (+1)
- order_patterns_suspicious is true (+1)
- referral_source_quality is "High" (+1)
**No Violation Score** (threshold >= 4):
- address_validity is true (+1)
- email_pattern_suspicious is false (+1)
- website_verified is true (+1)
- login_geographic_consistency is true (+1)
- payment_method_shared is false (+1)
- order_patterns_suspicious is false (+1)
Determine the violation type based on the HIGHEST score that meets its threshold. If multiple scores meet their thresholds, choose the highest score. If tied, use priority: Abusive Account Creation > Misleading Ad Copy > Personal Orders. If no score meets any threshold, the result is INCONCLUSIVE.
Apply the enforcement action:
- Abusive Account Creation → "Account Closure"
- Misleading Ad Copy → "Account Closure"
- Personal Orders (Related) → "No Action"
- No Violation → "No Action"
- INCONCLUSIVE → "Inconclusive"
在 GitHub 查看