When records match to more than one cluster, Ab Initio automatically logs alternatives and flags potential data quality issues for review.
Many matching applications require clustering — the grouping of similar records together. However, when clustering data, not all membership decisions are certain. Incomplete or conflicting information can create scenarios where a record matches multiple clusters. When this happens, Ab Initio’s clustering algorithm will put the record in one cluster and log the alternatives. The existence of alternatives can later be used to flag clusters for manual review, drawing attention to data quality issues that may hinder conclusive decisions. This highlights issues that may impact the accuracy of clustering rather than ignoring them or brushing them under the carpet.
How do you get your anti–money laundering software out of the spin cycle?
A large European bank trying to prevent money laundering was using software that was trapped in a never-ending spin cycle. Even with a relatively small number of names, checking those names against sanction lists was tremendously difficult. This did not make the regulators happy.
Using industry-standard software to perform their matches, the bank needed several months just to get their basic rules up and running. Then, the bank couldn’t explain the matches that were being made. Even worse, the software experts couldn’t explain the matches. Attempts to customize the rules were unsuccessful; eventually, the bank learned that they needed additional products from that vendor to get all the features they needed.
The bank decided to go with another vendor. This new solution also took several months to implement, produced a massive number of false positives, and took days to process only a few thousand records. As before, it was impossible to determine why the software had generated those matches.
At this point, the bank called on Ab Initio.
Using Ab Initio’s intuitive spreadsheet-like interface for rule development, it took only a couple days to implement the initial set of rules. Once those rules were in place, in less than an hour Ab Initio processed the same set of records that had taken the previous solution days.
When presented with a list of over 100,000 potential matches, Ab Initio software determined in less than an hour that almost half of the matches were false positives and provided a clear diagnostic trail showing why it had obtained those results. For each entry flagged as a correct match or as a false positive, it was easy to understand why the software had made that decision.
Thanks to Ab Initio, the bank finally got their anti–money laundering process out of the spin cycle.