I have noticed that the SAP kernel running on HANA has more possibilities for fuzzy search and/or linguistic search. The example that I read is, that cafe, café, Cafe and Café are very similar.
This is not surprising. After all, if the database is in memory, such things are more feasible than with a classic rdbms - in addition, some of the database logic is moved from the application layer to the database layer.
It opens new possibilities in master data. When a user attempts to create a new customer, it would be much easier to detect that a customer with a similar name or at a similar address already exists.
I am wondering whether a workbench exists to create such similarity rules. After all, such rule-sets are very context-specific. Similarities between Dutch street names, for example, would not work well in a Norwegian context.
I think I have seen such a workbench for BODS (Business Objects Data Services), but that's not HANA - different target.