Privacy & Security Classification

The EU has set a global benchmark for data protection through the GDPR and established a very high standard for security classification with the NIS2 Directive.

Data & More has spent more than nine years becoming a market leader in privacy classification and over three years developing advanced security classification capabilities. This work has been carried out in close collaboration with clients’ legal teams across more than 28 countries, as well as with leading academics specializing in rare event classification. Rare event classification is a discipline focused on accurately identifying extremely uncommon signals—finding the needle in the haystack without collecting too much irrelevant data.

The result is a Multi-Factor Classification approach that combines multiple technologies. These range from basic token-based techniques such as regular expressions and dictionaries to advanced semantic and contextual analysis, graph-based methods, and multiple AI-driven models.

Our quality target allows at most 5 false positives per 100 documents.
On average, sensitive personally information (SPI) occurs in 0.5% of documents, meaning 1 document out of 200 contains SPI. In a sample of 20.000 documents, this corresponds to approximately 95 true SPI documents and 5 false positives.

To meet the quality target, false positives must therefore be limited to 0.025% of all positive classifications, which results in a classification precision of 99.975%.
In practical terms, this means our toolbox identifies rare PII instances with extremely high accuracy, despite the very low underlying signal density.

For more information on privacy and security classification please review our support site HERE.

This is an AI generated video