Noise words are common words within a company or organisation name which will be ignored when the matching process is taking place on the screening system, so that key name words will be the focus. Most screening systems allow you to configure a set of noise words, or at least include a preconfigured set. These noise words are removed from names prior to entering the matching system so are not considered by the rules and algorithms.
The purpose of identification and removal of noise words is to help refine the matching process by eliminating data from the names, which is less important in identifying the uniqueness of the name. These types of words may be left out of the sanctions list name or the customer name, so in theory, removal in the matching process will enable the scoring of alerts, to be more balanced.
For example:
Customer Name versus List Name
‘RGI International’ versus ‘RGI’
If ‘international’ was not considered a noise word the likelihood is that the match between these two names could fall below your alerting threshold and would not generate an alert for review. If ‘international’ was a noise word it would be ignored within the matching process and the system would see RGI versus RGI and would likely generate a high scoring match.
A critical requirement for use of noise words is to ensure that the noise words are removed equally from the sanctions lists names and your customer names so that there is a balanced view.
For example:
‘RGI International’ versus ‘RGI International’
If the system considered international a noise word for the list names but not for the customer names, the match being reviewed would be:
‘RGI International’ versus ‘RGI’ Again the same issue as above can arise whereby this might not be identified as a valid match within your screening threshold.
Determining what to use as a noise word will require thought. Most system vendors will include a default list which you may be able to amend internally or where you make a business case and the vendor will add the word to the default list for all clients. In particular, where the list is more generic for all customers, a full review should be completed with testing where the noise words functionality is switched off and then switched on to see the difference in the results.
The articles to follow will look at how to decide which words to include in your noise words list and the potential pitfalls with using noise words.
If you would like to learn more about this, and how SQA Consulting may assist you in such needs please contact us.
