There were several questions in this thread and other OFAC-related threads about OFAC software. Innovative Systems helps make BankersOnline possible by advertising with us. I asked for their expertise in responding to these software-related inquiries. Their helpful response is below. An article they've contributed on "How to Select OFAC Software" will also be up on the site beginning Monday.
The following info is from Chuck Schardong,
Senior Product Manager, Innovative Systems, Inc.:
Many of the matching concerns and issues expressed in the BankersOnline OFAC threads seem to revolve around three key areas: quality, reliability and the detail of the results being generated from the various OFAC software systems. All are very important concerns.
Matching Methodologies
There are several matching methodologies used in the marketplace today by OFAC solution providers -- including basic matching algorithms such as string-comparison, key-based and weighted-field matching technologies. In my opinion, the issues being discussed are consistent with known constraints inherent to these matching methodologies.
For example, the Miranda=iran case is a potential problem of string-comparison algorithms, because this technique cannot provide the true context for the data being matched. Likewise, not being able to specifically demonstrate why two records matched within a given score/filter, such as the boy scout records, is a potential problem with weighted-field technologies. The logic behind these technologies tend to promote an increase in false-positive match rates which, in turn, creates additional manual review work for the compliance staff or brings to question the reliability and capability of the products.
Parsing Data Also Important
It should also be noted that matching shortcomings may be compounded by the parsing routines used by the basic matching algorithms. Parsing routines are the foundation for choosing the data elements which will be compared by the matching engines. Since match results can be impacted by poor data quality, parsing routines should be able to address existing issues in the data to more accurately choose the most appropriate data for matching. Parsers should also be able to decipher and ignore noise words in the customer data -- for example, relationship terms such as UGMA, MINOR, TRUSTEE and ITF, as well as name appendages such as JR, SR and PHD - that could impact match results. Minimizing the impacts of parser-related errors can greatly reduce the amount of false-positive matches.
How to Maximize Matching Accuracy
The key to maximizing the accuracy, reliability and usability of matching products is to combine a more sophisticated, heuristic matching algorithm with knowledge bases of words and patterns that guide the parsing process. This combination permits accurate matching in spite of misspellings, transpositions, missing information, and other variations and anomalies. It will prevent, for example, an individual's name from being incorrectly compared to a street or country.
What to Look For
For accurate, reliable OFAC matching - or any customer data matching, for that matter - the software you select should include:
· Matching methodology that minimizes or eliminates false positive matches without causing missed matches
· Output that clearly identifies how and why records were matched
· Flexibility to match on both name and address, as well as name only, or address only
· Opportunity for users to specify their own unique matching criteria and thresholds.
· Adequate parsing capability - because matching technologies are often only as good as the parsers which help determine the fields to be compared
· Accurate and consistent matching in both online and batch environments
· And, to better justify the cost from a business standpoint, the software should be able to be used not just for OFAC compliance, but wherever customer matching is required
For more information on OFAC compliance, or for a copy of Innovative's White Paper on Data Matching Architecture which better outlines the advantages and disadvantages of the available matching methodologies, please visit our web site or contact us at 800.622.6390.