Table of Contents

Search

  1. Preface
  2. Introduction
  3. Definition File Overview
  4. Customization Steps
  5. Service Group Definition
  6. Algorithm Definition
  7. Edit-list Definition
  8. Matching Scheme Definition

Service Group Definition and Customization Guide

Service Group Definition and Customization Guide

Tips on Developing a Matching Scheme

Tips on Developing a Matching Scheme

Schemes can be developed using either the Matching Scheme Definition file or by using the run-time DEBUG Service. This section will only describe the creation of a Scheme using the Definition file. See the Debug section in the
APPLICATION REFERENCE FOR SSA-NAME3 SERVICE GROUPS
guide for more information on the use of the DEBUG Service.

Building a Multi-Method Scheme

This section describes how to develop a simple Scheme to match one field from two records, and then how to expand the Scheme to add more fields.
To match a single field in two records a Scheme similar to the following example would be used,
COPY SSASCRM * DEFINEMETHOD=MPERS,EP=N3SCM,ALGNAME=PERSON DEFINE * SCHEMENAME=FULLNAME METHODNAME=MPERS,GOPT=(LENGTH*50) FIELDOFFSET=0 * SCHEMEEND END
This defines a Scheme called
FULLNAME
which uses Method
N3SCL
to compare the person’s full name in the search and file records. The underlined words are user-defined. The Algorithm name (
ALGNAME=
) must be the name of an authorized Algorithm in the Service Group. The field length (
LENGTH*
) must be the same as the
NAME-LENGTH
field in that Algorithm and must be the length of the fields passed to the Scheme for matching.
Now let us assume that the person’s name can be formatted into family-name and given names and we want the person’s family name to be more important than the given names. The Scheme could be modified to use two Methods, one will compare the family name with a higher weighting and the other will compare the given names with a lesser weighting;
COPY SSASCRM * DEFINE METHOD=MPERS,EP=N3SCM,ALGNAME=PERSON DEFINE * SCHEMENAME=FULLNAME METHODNAME=MPERS,WEIGHT=3,GOPT=(LENGTH*20) FIELDOFFSET=0 METHODNAME=MPERS,WEIGHT=1,GOPT=(LENGTH*30) FIELDOFFSET=20 * SCHEMEEND END
Using the above example, the following diagram shows how a result is achieved when matching the following two records:
GRAY ROBERT JAMES GRAY ROBERT MICHAEL
For a full description of the above formula, refer to section Score Calculation by a Scheme.
Note that Scores are usually rounded up, in this case 88% would be returned. (If the same weight had been given to both fields, a Score of 75% would have resulted).
Finally we decide that – actually – the most contributing piece of information to the match is the person’s address (we are saying, if the address is wrong, we are not interested in this person). So, another Method is added to the scheme to compare address and give it a higher weighting;
COPY SSASCRM * DEFINE METHOD=MPERS,EP=N3SCM,ALGNAME=PERSON DEFINE METHOD=MSTRT,EP=N3SCM,ALGNAME=STREET DEFINE * SCHEMENAME=HOUSHOLD METHODNAME=MPERS,WEIGHT=3,GOPT=(LENGTH*20) FIELDOFFSET=0 METHODNAME=MPERS,WEIGHT=1,GOPT=(LENGTH*30) FIELDOFFSET=20 METHOD NAME=LSTRT,WEIGHT=4,GOPT=(LENGTH*80), LOPT=(CONC) FIELD OFFSET=50 * SCHEMEEND END
This definition adds Method definition
MTRT
which acts on the "address" field in the search & file records (positions 50-129) and again uses the N3SCM matching method. This results in a Scheme with an extreme bias towards those records with the correct address, as only such records have a chance of scoring above 50%. The following diagram shows how a result is achieved when matching the following two records:
The Score of 092 for the address could be modified by using different method options and values.
The actual returned Score would be 090.
This example shows the sort of approach which might be used for a house-holding application.

0 COMMENTS

We’d like to hear from you!