ITSV manages 500 billions of data records in relational databases for Austrian social securities, which gives them a powerful dataset. Apart from having to encrypt and pseudonymizise the data, they struggled with the biggest challenge of data warehouses world wide at the moment: How to use AI methods to check the normality of their data. But keeping up with over 9000 rules in their model proved a challenge. Hear in this presentation, how throwing out the rules improved the success of their approach massively.

• No more maintenance of ranges required
• Training in the background to keep attention where it matters
• No case of anormal data since full implementation