The IBM Entity Analytic Solutions Name Manager name matching algorithm
is designed to yield culturally aware name scoring.
The IBM Entity Analytic Solutions Name Manager name matching algorithm
is used in conjunction with the Name Comparator 1.0 or Name Comparator 2.0
name matching algorithms. IBM Entity Analytic Solutions Name
Manager groups names into three categories which it then compares and scores:
- Given name (first name and middle – or all words but the last name)
- Surname (last name)
- Full name (all words)
Configuration considerations - confidence threshold guidelines
These
three scoring confidence threshold parameters allow you to tune name matching
for specific resolution rules to meet your name matching requirements. Confidence
thresholds are decimal-based, from 0.0-1.0, with 0.0 being the lowest score
and 1.0 being the highest score. The higher the score in a category, the closer
the names matched in that category.
When using the IBM Entity Analytic
Solutions Name Manager name matching algorithm, if either the Name Comparator
or the IBM Entity Analytic Solutions Name Manager thresholds are met, the
confirmation or denial will be made. Only one set of the thresholds need to
be met.
- Name Manager Full Name Confidence Threshold:
- Based on the 0.0-1.0 score, here are guidelines to assist you in determining
the level of matching the Full Name score threshold:
- 1.0 = exact or near exact match
- 0.9 = close match (suitable for name and DOB resolution)
- 0.8 = good match (suitable for most resolution rules)
- 0.7 = average match (suitable when unique numbers are also present)
- Below 0.7 = not suitable for matching
- Name Manager Given Name Confidence Threshold:
- Based on the 0-100 score, here are guidelines to assist you in determining
the level of matching the Full Name score threshold:
- 1.0 = exact or near exact match
- 0.9 = very good match (may indicate given-surname swap)
- 0.8 = good match
- 0.7 = average match
- Below 0.7 = not suitable for matching; may be useful when combined with
Surname or Full Name to ensure some similarity
- Name Manager Surname Confidence Threshold:
- Based on the 0-100 score, here are guidelines to assist you in determining
the level of matching the surname score threshold:
- 1.0 = exact match
- 0.9 = very good match (may indicate given-surname swap)
- 0.8 = good match
- 0.7 = average match
- Below 0.7 = not suitable for matching; may be useful when combined with
Given Name or Full Name to ensure some similarity