Compliance requires name matching: for instance, to match customers against sanctions and PEP lists. Unfortunately, the name of a single person can be spelled in many different ways. Hence, a decision has to be made whether matches of only very similar names should be verified,or matches of less similar names. In the former case, we may miss relevant matches and in the latter, we may get too many matches.

risk matching

Considering all relevant matches, the portion of matches eventually found by the name matching is called recall. A high recall minimizes the risk of missing relevant matches. In all matches found by the name matching, the portion of relevant matches is called precision. A high precision minimizes the risk of too many verifications.

risk matching2

A 100% recall and precision would be desirable, but is impossible since recall and precision are mutually conflicting goals. Hence, a balance between recall and precision implies a trade-off between two different risks. Finding the optimal trade-off, or equivalently finding the optimal-name matching configuration, is difficult, since it depends on different aspects, such as business strategy and risk aversion.

Complete Revision of the Federal Data Protection Act

The complete revision's draft of the Federal Data Protection Act is currently in political consultation. Data Protection is to be increased by giving people more control over their private data as well as reinforcing transparancy regarding the handling of confidential data.

Links: draft, report

Eurospider Information Technology AG
Schaffhauserstrasse 18
8006 Zürich