Since “second” and “row” in the source2 are glued together, Power BI will treat it as an entire text part. But for the “SECOND row”, they will be treated as two text parts: SECOND and row. That’s why they are not matched.
Here is a sample can help you understand the working principal.
If you would like to make the “SECOND row” match with “secondrow”, you could merge like this: delete 1 from Similarity threshold(optional) input box and tick Match by combining text parts.
Match by combining text parts option will try to combine text parts such as “Micro” and “soft” into “Microsoft” to find matches. But the 1 Similarity will only allow exact matches which will prevent the match between “SECOND row” and “secondrow”.