About Fuzzy Logic Scripts

Fuzzy logic scripts contain an algorithm that includes logic for exact string matching and fuzzy string matching. After you create and save a fuzzy logic script, you can use the Classifier node in Cognitive Analytics to pass values into the script and to receive a target value prediction out of the script.

To function properly, fuzzy logic scripts require the following:

A reference to a standard list that represents standard values for the script's prediction.
An input of a collection of words, such as a group of fields identified in a query, that are processed through the script's algorithm.
References to standard lists for text cleaning.

When executed, a fuzzy logic script:

Concatenates all the input words into a single set and cleans the text.
Identifies exact matches from the standard list of standard values.
Identifies fuzzy matches from the standard list of standard values.

If neither an exact nor a fuzzy match is found, a blank prediction is provided and a score of zero is returned.

If one match is found, the matched word is returned as the prediction along with the prediction's score.

If more than one match is found, the following criteria is applied:

The matched word with the highest score is returned as the prediction along with the prediction's score.
If there is a tie between matches with the highest score, the matched word with the highest weight per the standard list of standard values is returned as the prediction along with the prediction's score.
If there is a tie between both the score and the weight of the matches, the first matched word in alphabetical order is returned as the prediction along with the prediction's score.