Table of Contents

Search

  1. Preface
  2. Introduction
  3. Major Concepts
  4. Prototyping
  5. The Design Issues
  6. Standard Population Choices
  7. Customer Identification Systems
  8. Identity Screening Systems
  9. Fraud and Intelligence Systems
  10. Marketing Systems

Best Practices Guide

Best Practices Guide

Search Levels

Search Levels

Using Standard Populations, an application may be set up to search on Person Names, Organization Names and Addresses using four different Search Levels:
  • Typical;
  • Exhaustive;
  • Narrow;
  • Extreme.
The choice of Search Level is passed to the SSA-NAME3 "get ranges" function directly by the user’s application.
It is good practice to test using different Search Levels on real production data and volumes to measure both the response time and the reliability differences.

Typical

A Typical search level for most applications will provide a practical balance between quality and response time. It should be used in typical online or batch transaction searches. It is the default if no Search Level is specified.
For
Person_Name
searches, it is designed to find common, but not extreme, error and variation including cases where initials are present instead of full given names and where the initial of a name has changed due to the internal rules applied.
For
Organization_Name
searches, it is designed to find common, but not extreme, error and variation including instances of word concatenation.
For
Address_Part1
searches, it is designed to find common, but not extreme, error and variation.

Exhaustive

An Exhaustive search level is provided for applications that have an increased risk associated with missing a match, where data quality is a concern or where data volumes are low enough to make it the default search. It increases the number of candidates returned and consequently response times may be extended. An Exhaustive search will occasionally find matches that a Typical search misses, however, these will generally be where there is more extreme error and variation.
For
Person_Name
searches, it is designed to find more error and variation than a Typical search, especially where there is extreme spelling error in the family or middle names.
For
Organization_Name
searches, it is designed to find more error and variation than a Typical search, especially where there is extreme spelling error in the major word or trailing words.
For
Address_Part1
searches, it is designed to find more error and variation than a Typical search, especially where there are more cases of missing words, extra words or sequence differences.

Narrow

A Narrow search level compromises on completeness of search in favor of faster and more direct answers. It may be an option in search applications that do not have a high risk associated with missing a match, require very tight levels of matching, or where data volumes are extreme and response time is a critical factor.
For
Person_Name
searches, it is designed to find the very common error and variation including cases where initials are present instead of full given names.
For
Organization_Name
searches, it is designed to find the very common error and variation and primarily where the words are in a stable order.
For
Address_Part1
searches, it is designed to find the very common error and variation and primarily where the tokens are in a stable order.

Extreme

An Extreme search level uses every possibility to discover a candidate match; consequently response times may be extended. It is provided for applications that have a critical need to find a match if one is present in the database, despite the error and variation.
An Extreme search may only occasionally find matches that an Exhaustive search misses, however, because the risk is very high, every possible match is deemed important.
The types of candidates returned for all Field types is the same when using an Extreme search. Extreme spelling error is picked up in names or addresses with two or more words or tokens.

0 COMMENTS

We’d like to hear from you!