RiskDetector
The Risk Detector flags various types of possibly fake and dummy data in person records.
How it works
The methods used to classify input data are divided into positive validation and negative checks.
Positive Validation
Person's Name
Parse tree validation, term and culture confidence consideration.
Physical Address
Parse validation, existence confirmation, geocoding. Verifies street address within postal code and place name.
Telephone Number
Correct parsing for location, area code existence verification.
Email Address
Syntax validation, domain existence, MX records verification, domain classification into risk scores.
Negative Checks
AI Neural Network
Trained on classified good and bad data.
Dictionary Lookups
Manually categorized terms for known patterns.
Rule-Based Computing
Detection for random typing and nonsense input.
Types of Invalid Input
Common Patterns
- • Random typing: "asdf asdf"
- • Placeholders: "John Doe", "Anytown"
- • Famous entities: "James Bond", "Barack Obama"
- • Test data: "firstname lastname", "test1 test2"
Disguised Input
- • Padding: "XXXJohnXXX"
- • Stutter typing: "Petttttttterson"
- • Spaced typing: "P e t e r M i l l e r"
Input/Output
Input
Uses common Ontology as input objects to REST API. Flexible data feeding:
- • Person's name (full name or separated fields)
- • Telephone numbers
- • Email addresses
- • Physical addresses
Output
Returns comprehensive risk assessment:
- • Overall risk score: Range -1 to +1
- • Score > 0: Risk detected
- • Score = 0: Neutral (nothing detected)
- • Score < 0: Genuine record
- • Detailed information about each detected risk
Ready to Detect Risk?
Try our live demo or get in touch to integrate RiskDetector into your workflow.