Verato Auto-Steward is designed to answer the question “Do these two sets of identity data actually represent the same person?” Auto-Steward does this through a SOAP web service in which the customer provides two sets of identity data as a Request. Verato sends a Response indicating whether the two sets of identity data are the same person, are NOT the same person, or if it cannot be determined. With this response, Verato also includes appropriate supporting evidence and metadata about Auto-Steward’s match determination.
In evaluating a request and forming a correct response, Verato Auto-Steward uses proprietary technology and processes that employ some specific terminology and key concepts that are defined below:
Verato compares your customer identity records to Verato’s reference database of US identities known as the Reference Database. The Reference Database is a comprehensive, authoritative reference database of US identities with over 300 million identities. By comparing your customer records to the Reference Database, Verato can find matches between your customer records that may not be apparent based only on comparing your customer records. For example, you might have two customer records, one for Sally Smith in Chicago, IL and another for Sally Jones in Detroit, MI. Based on that information, the records do not seem to be the same person with a high degree of confidence. But the Reference Database is aware of the longitudinal history of Sally’s information—so the Reference Database knows that Sally Smith changed her last name to Jones and moved to Detroit—she is in fact the same person, and Verato is able to match the customer records using the Reference Database data as a reference.
When a pair match request is received by Verato Auto-Steward, each input identity is compared against Verato’s Reference Database to determine if there is a matching reference identity. If the input identity matches a reference identity at a score of .8 or higher, this is considered a referential match. If both input identities referentially match to the same Reference Database identity, then the two input identities are considered a referential match to each other – they are deemed to represent the same individual. If each input identity referentially matches to a different Reference Database identity, then Auto-Steward considers the input identities to be a referential non-match (they represent two different people).
If one or both input identities do not match to a Reference Database identity, the two input identities are compared directly against each other. If the input identities match against each other at a score of .8 or higher, they are considered a head-to-head match.
The PairMatchRequest is the web service request (in XML format) your application sends to Verato Auto-Steward. This PairMatchRequest element is made up of 3 key sub-elements.
- RequestId – this is a string that clients can use to provide a unique ID or string of information to track each web service request. The RequestId should uniquely identify each web service request. The RequestId is returned in the web service response to associate the response to the request. Many Verato clients use the RequestId as a mechanism to associate each web service request back to the data stewardship tasks in their MPI/MDM system.
- Identity1 – this element is made up of the personally identifiable information (PII) for the first of the two input identities.
- Identity2 – this element is made up of the PII for the second of the two input identities.
Attributes and Attribute Clusters
Each Identity you include in a PairMatchRequest, and each reference identity in the Reference Database is comprised of multiple demographic data values. These values are known as Attributes. Some attribute values are logically grouped together into Attribute Clusters. For example, the attributes of first name, middle name, and last name are grouped together into an attribute cluster known as Name.
Attributes exist in both the PairMatchRequest as well as the Reference Database reference identity data model. They have pre-defined data structures and are used in the matching process to determine whether two identities are the same or not. The Attribute Clusters and Attributes most commonly used within the PairMatchRequest are:
|Street Line 1
|Street Line 2
The PairMatchResponse is the web service response (in XML format) Verato provides your application in answer to a PairMatchRequest. The specific elements of PairMatchResponse are described in detail (with examples) in section 3. The section immediately below provides a brief definition of some key elements in the PairMatchResponse:
- SameIdentity – this is a Y/N/U code that represents the answer to the question, “Do two sets of identity data represent the same person or not?”.
- If SameIdentity=Y, then the Auto-Steward service is asserting that the two identities represent the same person.
- If SameIdentity=N, then the Auto-Steward service is asserting that the two identities represent two different people definitively.
- If SameIdentity=U, then the Auto-Steward service is asserting that there is still ambiguity and the two identities are not deemed to be definitively the same or not – the outcome is Uncertain.
In each PairMatchResponse there are three separate SameIdentity elements. The first provides Auto-Steward’s answer to the primary question, “Do these input identities represent the same person or not?”. Later in the PairMatchResponse, each input identity receives a SameIdentity value representing the input identity’s comparison against the Reference Database.
- Memo – this is a text string explaining how Verato arrived at the primary SameIdentity decision (whether the two input identities represent the same person). For example, if each of the two inputs was a match to the same reference identity in Verato’s Reference Database, then the memo string would indicate something like this: “Identity1 and Identity2 matched the same commercial identity and therefore are the same identity.”
- NormalizedScore – this is the score, on a scale from 0.0 to 1.0, that Verato assigns to an overall match decision. Two identities are considered a match when they have a score of 0.8 or higher.
- Enhancement – Enhancements are modifications to the NormalizedScore that may occur when comparing an input identity to a Reference Database reference identity. Enhancements can either be bonuses that increase the match score due to uniqueness of the identity or penalties that decrease the match score when certain conditions are detected that indicate for example a potential parent/child pattern or a twin sibling pattern.
- HeadToHeadMatch – this describes the overall SameIdentity response for the pair.
- If HeadToHeadMatch=true, Auto-Steward based the final SameIdentity decision on a head-to-head match of the two input records.
- If HeadToHeadMatch=false, Auto-Steward based the final SameIdentity decision on matches of the input identities to the Reference Database.