Can you please provide some input on best practices when comparing string of text that aren’t an exact match that we can use when comparing names in the requests and application?
For example :
If a submitter has mentioned a name ‘Billy’ as a first name and ‘Jackson’ as a Last name and we are comparing the name in request and in other application where that customer is already present but, his name is not an exact match.
So, is there any way that BOT can read the fuzzy names which are not the exact match?
Uipath does but unfortunately your application may have not fuzzy search. So what you are asking is not straight forward. There is one extreme resort, because Uipath has fuzzy search what you can do is search just with the last name in your application and you will get multiple results. extract the result set as data table and then for each row do a Fuzzy search on the first name and then you can proceed with the highest score. Download this package. Uipathteam.string.activities
Activity :Levenshtein algorithm
Advanced Disclaimer: There is a no ML or NLP in levenshtein algorithm so it will give very low confidence score ( <40) in case “William” was entered as “Billy”. A score of 70 and above is healthy . in some cases >50 may be acceptable . But there is always risk associated with the any score that is not 100%
Using that algorithm, I got a match percentage of 7.68 something for original string “Previous Deposit” for the OCR output “preV1ous epos1t”. Here the letter “D” comes as space character and “i” as “1”. But, length wise both are same , also since its a ocr output it won’t be much accurate right. So, Now, how do I get this ocr output string? Thanks in advance
I did levshtein comparison and the results were very positive for the 2 strings you’ve mentioned. The uipath string activity has levenshtein for string comparison and not for a string-Find. So you will have to write some steps to find if something fuzzily matching PreviousDeposit is found anywhere.