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

Weights

Weights

When using a scheme with two or more Methods it is sometimes appropriate to apply different emphasis on the results from each Method. For example in a Scheme that scores surnames with one Method and given names with another, it may be more important for the surname to match than the given names. In this situation the scheme can apply a factor to the surname Method’s result. This factor is called a weight.

Score Calculation by a Scheme

As mentioned, when more than one Method is used the Scheme has to combine the Scores from all the Methods into a Score for the record. This is done by calculating a weighted sum of the Scores. This technique involves multiplying each separate Score by its weight then dividing the total by the total weight. For example, if the record has a name at offset 0 in the record and a date at offset 50 you may define a scheme as follows:
COPY SSASCRM MODULE N3MAZZ DEFINE METHOD=NAME1,EP=N3SCL,ALGNAME=PERSON DEFINE METHOD=DATE,EP=N3SCD DEFINE SCHEME NAME=SCHEME1 METHOD NAME=NAME1,WEIGHT=75,GOPT=(LENGTH*50) FIELD OFFSET=0CLIENT NAME AT START OF RECORD METHOD NAME=DATE,WEIGHT=25,GOPT=(LENGTH*50) FIELD OFFSET=50DATE AT OFFSET 50 SCHEME END END
The Scheme is called
SCHEME1
and uses a name matching Method (
N3SCL
) for the name with a weight of 75 and a date matching Method (N3SCD) for the date with a weight of 25.
The total Score returned from this scheme is:
The total weight does not have to be 100, any total is OK. If, in the above example, the default weight of 1 was used for the name and the date (i.e. the
WEIGHT=
directive was not included in the above definition) then the total Score would have been: where the 2 in the above equation is just 1+1 (the total weight).

Weight Modification by a Method

A Method can further control the total Score by modifying its own weight. If, for example, the Scheme uses two methods, each with equal weight, then either method may decide that the data it finds in the record justifies a higher (or lower) weight.
Although the Method does not know what weight is assigned to it, it can return a weight-modifier which is a non-negative number that will be used to adjust the weight. Most Methods will return a weight-modifier of 100 which means "leave the weight as defined by the user in the Scheme definition".
A modifier of 200 will mean "double my weight" and a modifier of 50 will mean "halve my weight". A modifier of 0 will cause the method to have NO contribution to the total Score, which (if there are two Methods in a Scheme) means that the total Score is identical to the Score from the other Method.
This is the formula used for a Scheme with two methods. Let us use some notation:
  • W1
    = weight defined for first Method
  • S1
    = Score returned by first Method
  • M1
    = weight-modifier returned by first Method
  • W2
    = weight defined for second Method
  • S2
    = Score returned by second Method
  • M2
    = weight-modifier returned by second Method
This means that if a field is blank a Method can give it a Score of zero and a weight modifier of zero. In this way the Method causes the blank field to have no effect on the total Score.
Also if the field contains incomplete data (for example year and month only – no day of month) then the weight can be reduced thereby reducing the effect on the final total Score.
Finally if the field contains extra data then a weight modifier greater than 100% will increase its effect on the final total Score.
Example Weight Modifier Usage
Consider the following Scheme definition,
COPYSSASCRM DEFINEMETHOD=N,EP=N3SCL,ALGNAME=PERSON DEFINE SCHEMENAME=SCHEME1 METHODNAME=N,WEIGHT=75,GOPT=(LENGTH*50) FIELDOFFSET=0 METHODNAME=N,WEIGHT=25,GOPT=(LENGTH*50+NULLE) FIELDOFFSET=60 SCHEMEEND END
Two methods have been specified in this Scheme. Normally if there is valid or useful data in all the fields then the first Scheme will be given a weight of 75% and the second 25%. However, because the second Scheme had the
NULLE
option it will return a weight modifier of 0 if it detects invalid (or not present) data in one or both of the fields it matched. This would ensure that only the first Scheme contributed to the final Score.

0 COMMENTS

We’d like to hear from you!