如何在雪花中进行模糊匹配 / SQL

Question

How to do fuzzy match in Snowflake / SQL

Here is the business logic

The ABC Company INC, The north America, ABC (Those two should shows a match)

The 16K LLC, 16K LLC (Those two should shows a match) enter image description here I attached some test data. Thank so much guys!

Answer 1

Any matching attempt that treats string pairs like "The ABC Company INC" and "The north America, ABC" or "Preferred ABC Group" and "The Preferred Residences" as a match is probably going to give you many false positive matches, since in some of your examples there is only one word similar between the strings.

That said, Snowflake does provide a couple of functions that might help: EDITDISTANCE and JAROWINKLER_SIMILARITY .

EDITDISTANCE generates a number that represents the Levenshtein distance between two strings (basically the number of edits it would take to change one string into the other). A lower number indicates fewer edits needed and so potentially a closer match.

JAROWINKLER_SIMILARITY uses an algorithm to calculate a "similarity" score between 0 and 100 for two strings. A higher number indicates more similarity, 100 being an exact match.

You could use either or both of these functions to generate scores for each pair of strings and then decide on a threshold that best represents a match for your purposes.