Configuring TBAML for Different Scenarios

As previously mentioned, Oracle Financial Services TBAML includes clusters and matching rules that are suited to various screening requirements.

Tuning TBAML to match your policies should be undertaken carefully and under the supervision of a risk and compliance expert, with knowledge of your business requirements and the relevant legislation.

The following general points may be useful when tuning the behavior of TBAML:

• Some organizations use a zero-tolerance policy for individual name matches. Such a policy typically requires that all potential name matches must be manually reviewed, irrespective of the rest of the data associated with the record. If such a policy is in place, you must enable the conflict rules in individual matching.
• In general, using a loose clustering strategy will result in relatively few clusters, each containing many records. This reduces the potential for missing true matches, and increases the chance of false positives. It is also expensive in terms of processing requirements, as every record in the cluster must be directly compared with every other record in that cluster.
• Conversely, a tight clustering strategy will result in a relatively large number of clusters, each containing fewer records. This increases the potential for missing true matches, and decreases the chance of false positives. It also reduces the overall cost of the processing requirements, as relatively few direct comparisons are required.