What happens when the conversion error limit is reached in fuzzy matching tool?

[knozawa]

What happens when the conversion error limit is reached in fuzzy matching tool?

Does that mean there is no more processing available in fuzzy matching tool even if there still exists non-unicode characters for matching?

Sincerely,

knozawa

[JoeS · Accepted solution]

I think that message means that its stopped reporting the conversion errors, but not processing the data.

So you should be fine on the output, but just not seeing every record that has a conversion error.

Thanks

Joe

[kmgoins1991]

I am having the same error. I am also having over 50% of my matches on company name match at 100% that are not even close. 

For example, Active Pest Control matching with B&B Petroleum. 

The other 50% seem to match correctly. I have attached an image of the matching logic I am using.

Is the problem deriving from the error "unicode characters were not convertible"... and "Company: Field conversion error: Limit Reached"?

Thanks for the insight

[JoeS]

Hi @kmgoins1991

That does seem very odd.

Would you be able to attach the data (a sample would be fine) and the workflow.

I have tried with the two examples you have put and the same settings and don't get a match.

One thing to add however, is the key length of 4 is quite short, but it's not that as I set it up the same on my machine.

[kmgoins1991]

When I created a smaller sample size, with names showing a true match and those that shouldn't, it doesn't have the issue. Any idea what could be deriving this?

I have attached the workflow for your view, but am hesitant to provide any data.

Thanks

[JoeS]

Hi @kmgoins1991

I do understand that you don't want to upload any data.

Looking through your configuration I can't see anything obviously wrong.

In the fuzzy match tool you can output the generated keys. I'd recommend doing that and see what has been generated from each record as that may give some insight.

It's in the advanced options at the bottom of the configuration