A whole new way to look at Knowing Your Customer & AML, Socially.

In July 2016, ID+ v2.5 added new field validation models, reason codes and correlation scores that are providing incredible predictive power that feed into both the fraud and authenticity scores returned from an ID+ Digital Identity Check. The correlation score is a new and unique way to improve upon an identity verification process, and enable AML and KYC-type inspection.

CIP and KYC/AML in nut shell 

Until now, Customer Identification Programs (CIP) and Anti Money Laundering (AML) Due Diligence processes for non-documentary identity verification were focused on verifying personally identifiable information such as a customer’s Name, Address, Date Of Birth and federal or state issued government identifier such as a Social Security Number or Driver License number that is passed to Credit Bureaus and Gramm-Leach Bliley Act (GLBA) data broker agencies for PII comparison based validation.

Due to rash of data breaches in recent years, the information required to pass these online, non-documentary ID verification requirements can be purchased from the “dark web” for five dollars or less. Socure has endeavored to go beyond simple KYC PII comparison-based verification for AML, CIP and Customer Due Diligence by combining offline, online and social behavior information in real-time, in order to verify consumer identity and document verification in a more robust way, allowing enterprises to meet and exceed regulatory requirements to Know Your Customer.

Machine learning and big data analytics applied to KYC checks

In July 2016, ID+ v2.5 added machine learning models and correlation scores that are providing significant predictive power that feed into both the fraud and authenticity scores returned from  a Socure Digital Identity Check. This month, as part of our regular improvements we’re making to our ID+ solution for consumer verification, we’ve included a significant improvement to our correlation models.

NAPE, NAP, NAE and NPE Correlation Scores

Name, Address, Phone and E-mail are relevant identifiers often combined with other private information like date of birth (DOB) and National Identification Numbers [such as a US Social Security Number (SSN)] for today’s digitally active consumer.

With the combination of GLBA credit header data, online and social data, mobile network operator data and offline PII data to our platform, our data science team has come up with a unique PII correlation model that is able to quickly verify if there is something inconsistent with a given Identity being validated for AML/KYC purposes, in an empirical way. This new model goes beyond simply fuzzy matching and equality tests for PII, examining how the personas related to the PII are linked together in complex ways, helping to identify where there are anomalies related to the profiles being matched against the PII in real-time. With the introduction of the new correlation models, we are happy to announce the introduction of the following scores:

• Name_Address_Phone_Email (NAPE) — four dimensions of correlation in one score
• Name_Address_Phone (NAP) — three dimensions of correlation in one score
• Name_Address_Email (NAE) — three dimensions of correlation in one score
• Name_Phone_Email (NPE) — three dimensions of correlation in one score

KYC Field Validation (legacy)

For clients who have leveraged the legacy capability of ID+ checks from v2.0, our KYC Field Validation explained how well each API input field matched with the given name of a consumer.

For example, if a subject identity verification had an email that did not correlate with the supplied name of the identity, our KYC Field Validation score (a number between 0 and 1) might return a value < 0.5 for that email address. With our new NAPE correlation score, our platform goes beyond basic field validation to produce a correlation score across multiple dimensions of data in one score, avoiding the need to examine multiple Field Validation scores for the given input data. We expect to retire our legacy Field Validation scores in a future release in favor of the new NAPE correlation score.

FIELD VALIDATION CHANGES

The following is the new Field Validation structure which will be included in our JSON response:

{ “details”: [{    “fieldValidation”: { “address”: “0.99”, “dob”: “”, “name”: “0.73”, “email”: “0.99”, “phone”: “0.99” },  “correlationScores”: {“name_address_phone_email”: 0.99, “name_addess_phone”: 0.99   “name_address_email”: 0.99, “name_phone_email”: 0.99 }  } }

If any of the component data elements is not provided as input into our API – for example no phone number provided to ID+, then the respective scores with phone as a component will not be calculated; and therefore NAPE, NAP and NPE scores will be returned as NULL in this example.

MANUAL VERIFICATION PROCESS DETAILS

For know your customer audit and manual review justification purposes, we are introducing the following Informational and Risk reason codes that will fire for each type of correlation score:

• I708 – Name associated with address matches input name
• R705 – Name associated with address does not match input name
• I618 – Name associated with phone matches input name
• R608 – Name associated with phone does not match input name
• I556 – Registered name for email address matches input name
• R559 – Registered name for email address does not match input name

The informational (‘Ixxx’) reason codes are fired only if a definitive positive determination can be made using the data available from our data sources.  If a definitive positive determination cannot be made because there are anomalies, the risk (‘Rxxx’) reason code is fired.  If no data is available from our data sources, then neither of the reason codes will be fired.

Note that no changes are being made to existing field validation scores for name, address, email, DOB, and phone. These will be deprecated in a future release and Socure product management will provide advance notice before this event.