The Matchers flipper of a decision table holds matchers that compare values on two objects and produces a match score. The input values are generated by bind variables or data elements (as defined in the Match Criteria Data Elements topic here). The match score is used in the Rules result formula to calculate the final match score (as defined in the Match Criteria Rules topic here).
To review examples of how matching algorithm uses match scores to compare and link objects in a Golden Record Clerical Review Task List, refer to the Potential Duplicate Match Score Examples topic here.
Matchers can be used to enable or disable Rules. For example, a rule may specify that it is only relevant if the address match score is above 70, or if address match is 'True' (depending on the inner score threshold definition of the matcher). Many matchers allow a default threshold setting to determine what is considered 'True' or 'False' in a match rule condition.
Many matchers include default weights and metrics, allowing detailed calibration of the algorithm to specific datasets. Calibration is handled during match tuning (as defined in the Match Tuning topic here).
|
Matcher |
Matcher Type |
Object Type Allowed |
|---|---|---|
|
Business Function (here) |
General Purpose |
Entities Products |
|
Function (here) |
General Purpose |
Entities Products |
|
JavaScript Function (here) |
General Purpose |
Entities Products |
|
Address (here) |
Preconfigured |
Entities |
|
Email (here) |
Preconfigured |
Entities |
|
Machine Learning Matcher
|
Preconfigured |
Entities |
|
Organization Name (here) |
Preconfigured |
Entities |
|
Person Name (here) |
Preconfigured |
Entities |
|
Phone (here) |
Preconfigured |
Entities |
|
Words (here) |
Preconfigured |
Entities |