Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.
Originally posted by: mmarinelliNode Name: Levenshtein Distance Match
Node Category: Correlation
Use Case: Use this node to compare field values which don't match exactly but logically should match (e.g. "Recognize" and "Recognise"). May be used as alternative to Soundex matching also included in this library.
Tested on Release: 4.1.6
External Dependencies: None
Practical Usage Example: An example use of this node has been provided in the prototype examples graph which is released in conjunction with the prototype library. Please see that graph for a executable use case.*
Node Documentation:
Functional Description:
Compares field values from two sources using the Levenshtein distance string matching technique, producing two outputs for matches: "main" - all rows from the "main" input for which a match was found, "lookup" - all matched rows from the "lookup" input. The output contains two new fields: "TEST_INDEX" - index of the matched set, which can be used to associate the "main" and "lookup" matched sets, "TEST_VALUE" - ("lookup" output only) the computed Levenshtein string distance which produced the match. Matches have distances within the tolerance indicated by the Threshold node parameter.
Parameters:
-
FieldNameMain: Field from "main" input which will be compared to "lookup" input
-
FieldNameLookup: Field from "lookup" input which will be compared to "main" input
-
Threshold: Distance threshold for determination of a match
-
GroupBy: Expression used to define groups within which the matching will be performed. This narrows the computation set and thus affects performance. Default is the first character of the field value.
Known Limitations: None