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

N3SCS – Date Matching

N3SCS – Date Matching

This method allows for more date variations than
N3SCD
or
N3SCJ
.
It accepts two dates and tries an exact compare. If that does not succeed, and a date format is not specified, it will try to match the pair using a variety of predefined patterns.
It is controlled using the following options.
LOPT=(CLEANEDT)
The option turns on cleaning and editing rules for Date Matching (but not the major marker rules). To use this option, the
Method Definition
must specify the algorithm that uses the Edit-list containing any special rules. The type of Edit-list rules must be character rule definitions.
For example, consider matching on a Date of Birth field and the value "010101" is a null value that shouldn’t be considered. Then the following Edit-list rule can be defined:
*S >010101< *W ><
In the matching definition, the method definition would look like:
DEFINE METHOD=DATE,EP=N3SCS,ALGNAME=<AlgName>
With the scheme using:
LOPT=(CLEANEDT)
LOPT=(DDMMYY / MMDDYY / YYMMDD)
The order of the date parts being passed. Select only one. This option is only required if a high degree of control over range matching is needed. Note that even though these options do not explicitly refer to the century part, inclusion of the century is fully supported. It is highly recommended that the century be included in the dates being matched.

Pattern Matching Options

The N3SCS date matching routine differs from the N3SCD routine because it includes logic to cope with dates that are not in a set format (i.e. MM/DD/YY etc). The N3SCD date routine requires a format to be specified. While the N3SCJ routine also does not require a date format to be specified, the N3SCS routine provides more options than N3SCJ.
An LOPT date format may be specified with N3SCS, and if so, gives greater control over range matching.
However, if an LOPT format is not specified, and if two dates do not match well in their original form, the N3SCS routine will use a series of pre-defined patterns to try to match the two dates.
If a match is achieved using one of these predefined patterns, the score can be penalized using the following option:
  • OPTION SCORES
  • VALUE OPENALTY,[number]
Specifies the penalty to apply if a derived pattern is used to generate a score.

Range Matching Options

Range matching can be performed on any or all of the date parts. The date parts are "day", "month" and "year", where year can include century. For example, range matching could allow two dates to match if they are one day apart, or up to five years apart.
The options below provide control over range tolerances, scores and penalties.
To use the first set of range matching options below the format of the date MUST be specified (see
LOPT= DDMMYY/MMDDYY/YYMMDD
above). See the end of this subsection for options that do not require the date format to be specified.
In order to disable range matching, the following options must be set to 0:
RANGE, RANGEDD, RANGEMM, RANGEYY, RANGESCR, RANGESCX
.
OPTION SCORESVALUE RANGEOPT,{0/1/2/3}
Controls which date part(s) can participate in the range matching. Allow range matching on all date parts Allow range matching on Year only Allow range matching on Month only Allow range matching on Day only:
  • OPTION SCORES
  • VALUE RANGEDD,[number]
  • VALUE RANGEMM,[number]
  • VALUE RANGEYY,[number]
Specifies the size of the range to be used for the part(s) required. For example,
OPTION SCORES VALUE RANGEYY,5
will allow for a range of 5 years in the two dates. If a value of 0 is specified, the value used comes from the
RANGE
option (see below sub-section).
OPTION SCORES
VALUE RANGESCR,[number]
Specifies the score out of 100 that a part match should achieve if matched inside the specified range.
OPTION SCORES
VALUE RANGESCR,[number]
Specifies the score out of 100 that a part match should achieve if matched inside the specified range.
OPTION SCORES
VALUE RANGEPEN,[number]
If
RANGEOPT
is set to 0 (allow range matching on all parts), this option specifies a penalty to be applied to any successful range matches after the first. For example, if the following is defined:
OPTION SCORES VALUE RANGEOPT VALUE RANGEDD,1 VALUE RANGEYY,5
and the following two dates are matched:
10/05/1963 11/05/1960
both DD and YY parts will score according to the value specified with
RANGESCR
, however the YY match will be penalized by the
RANGEPEN
value because it is a secondary range match. The following range matching options do not require a date format to be specified.
OPTION SCORES
VALUE RANGESTR,{0/1}
Setting
RANGESTR
to 0 allows range matching to be applied to all parts, in or out of position, and regardless of the date format.
OPTION SCORES
VALUE RANGE, [number]
Specifies the range value to use on all parts when
RANGESTR
is set to 0.

Age Range Matching Options

It is possible to check that an age calculated from a date falls within a specified range. The following options are used to enable and control Age Range Matching:
  • OPTION SCORES
  • VALUE AGERANGE,[0|1]
  • VALUE AGERNPEN,[number]
  • VALUE AGERNDEF,[number]
The date must be placed in the file record and the allowed range must be placed in the search record. The range may be specified in the form
n-m
, where
n
and
m
specify the inclusive lower and upper limits of the range into which the age should fall. E.g 0-21 and 65-100 are both valid ranges. Alternatively, the range may be specified as a single numeric
n
, in which case the calculated age must match n exactly.
In the case of a repeating field, only the range in the first field is used. If an invalid range is detected (e.g not numeric) a response code of 40 is returned.
The format of the date in the file record should match the date format specified for the method. However, if the first part of the specified date is 4 digits then the date format is assumed to be YYMMDD.
The
AGERNPEN
value represents the amount by which the score should be reduced if the calculated age does not fall within the specified range. The default value for
AGERNPEN
is 50. Here are some examples which illustrate how
AGERANGE
works:
SearchRecord
FileRecord
CalculatedAge
AGERNPEN
Score
45-50
19051960
46
20
100
50-60
19051960
46
20
80
50-60
19051960
46
default
50
In the above examples, the date format is assumed to be
DDMMYYYY
.
AGERNDEF
This is the score to return if no usable date is detected. The default is 0.

Other Options

The following options control other aspects of the date comparison.
OPTION SCORES
VALUE EXACTSCR,[number]
If two dates do not match exactly, this option specifies the score to apply to each exactly matching part.
OPTION SCORES
VALUE EXACTSCY,[number]
Specifies the score out of 100 to give a match of a 4-digit year (CCYY) against a 2-digit year (YY). For example: 10/05/1903
11/05/03
OPTION SCORES
VALUE EXACTSCX,[number]
Specifies the score out of 100 to give a match of a 4-digit year (CCYY) against a 1-digit year (Y). For example: 10/5/1903
1053
OPTION SCORES
VALUE NOEXCPEN,[number]
Specifies the penalty if the dates have different numbers of date parts. For example: 01 Nov 2003 0111
In this case, the matches would be 01 vs 01 and Nov vs 11 giving 100%. In order to reduce this score, because the first date has three parts and the second only two, NOEXCPEN,2 would reduce the match to 98%.
OPTION SCORES
VALUE PENALTY,[number]
Specifies the penalty to apply to the score for out-of-order matching parts. For example: 10051903
19030510
OPTION SCORES
VALUE TRANSSCR,[number]
Specifies the score to apply to transposed digit matches. For example: 21/05/1963
12/05/1963
OPTION SCORES
VALUE TRANSYY,[number]
Specifies the score to apply to transposed digit matches in the YY part of a year (CCYY). The default (if
TRANSYY
is not defined) is 75. If set to 0, the
TRANSSCR
value will be used. For example: 12/05/1936
12/05/1963
Both
TRANSSCR
and
TRANSYY
must be set to 0 to turn transposed digit matching off.

0 COMMENTS

We’d like to hear from you!