Hello Experts;
I would like to ask you for suggestion in terms of string comparison in Oracle. Normally we have available two functions related to distance similarity: edit_distance_similarity and jaro_winkler_similarity but if we consider that words can have different order those methods (their algorithms) will provide to us unexpected results:
e.g. "ABCD GGGG" & "GGGG ABCD"
words are the same and the only difference is in their order but edit_distance_similarity will get us 12 and jaro_Winkler_similarity will get us 44 as a result what is small score assuming that 100 is exact match and 0 is mismatch. I've checked that and I couldn't find any standard method to handle this problem, so I assume that the only way is to create PL/SQL to divide string into several words separated by blanc spaces and then we can do comparison using standard methods word by words (at the end we can calculate final score as average score considering numer of rows that we have). Do you have any suggestions how to do that in more efficent way?
Regards,
Bolo