Guest article provided by Jeanne Le Garrec, BasisTech 

Everyone who has ever had to assess or verify another human being’s identity knows the struggle. The stress of missing a name that appears on a watchlist. The time wasted investigating false positives. The complexity and frustration of parsing foreign names, often in foreign scripts.

Currently, banks, fintechs, insurance companies, law enforcement agencies, and other institutions dealing in high-stakes identity verification use three methods to determine who’s who: fingerprints, facial recognition, and name screening.

Issues have arisen with the first two methods.

Fingerprints can change over time. Facial recognition is a reliable method of identity verification, but it can be costly and perceived as intrusive.

That leaves name screening.

Name screening is both an effective and cost-effective method of identity verification. But not all name screening technologies are created equal. If you plan to implement or upgrade your organization’s name screening technology, make sure you invest in software that:

1. Uses AI for fuzzy matching

Fuzzy logic is a computing approach that improves upon Boolean processes by considering degrees of truth. In name screening, fuzzy logic is most commonly used to account for typos, transpositions, and other name differences and errors. It identifies similar but non-identical pieces of text appearing in separate records. It then ranks the likelihood of these similar pieces of text being a match.

This capability is important to name screening for a variety of reasons. Perhaps the most important of these is the different name components and name orders used in different datasets. A person’s name may be recorded using a first and last name (“Harold Jones”); a first, middle, and last name, (“Harold Andrew Jones”); a first name, middle initial and last name, (“Harold A. Jones”). Any one of these scenarios can be reversed by placing the last name first: “Jones, Harold Andrew” and so on. Boolean processes typically don’t consider disordered names a match — fuzzy matching does.

2. Tunes easily The identity verification field is always evolving. Criminals get smarter, new regulations are implemented to combat criminal activity, and organizations debut new processes to comply with those regulations. You need a name-screening solution that easily adapts to a changing environment. Your name screening solution should be capable of being tuned and updated easily, without significant time investment or production disruption.

3. Provides explainable scores Every organization and screening officer is accountable for the choices they make. Why did you permit this bank customer to onboard? Why did you allow that traveler to enter the county? Sometimes, regulators and others will want to know how the name-screening system scored the match as it did, and which AI models were used in the scoring process. Because of this, you should consider investing in a name screening system that offers “explainable AI.” Explainable AI is a set of methods and processes that enables users to better understand what AI is doing, on what data it’s basing its decisions, and how it calculated match scores.

4. Empowers you to adjust scoring parameters Understanding your name screening software’s scoring system will help you better understand why two names have been deemed a “match” or a “mismatch.” Unhappy with the decisions your system has made? The best name screening technologies allow you to track the “matches” that you would consider a mismatch, track the “mismatches” you would consider a match, and adjust scoring parameters accordingly.

5. Allows for benchmarking and testing You need the ability to benchmark and test your name screening solution in a specific sandbox, or with an independent sandbox provider. If a software provider will not allow you to test name screening software independently, consider it a red flag. Your solution should be tested before implementation, and whenever significant tuning or parameter adjustments have taken place.

Identity verification is inherently complicated, but the right name screening technology can ease the process. If you need to implement or improve name screening technology, make sure to take into account the criteria discussed above. To learn more, visit