We are pleased to announce Screena API version 2.1.17, a service pack that introduces key enhancements and bug fixes, ensuring more accurate and efficient screening of unstructured data fields.
Stopword detection in narratives: We have fixed an issue where stopwords positioned at the beginning of a narrative weren't detected correctly, leading to unnecessary alerts.
Geo matching from narratives: We have resolved an issue where the addresses object within sourceData occasionally returned a null value when matches on high-risk locations were identified through narratives. Now, the addresses object consistently provides the fullAddress attribute, which is set with the exact source element from the narrative that triggered the match. This enhancement leads to improved explainability. Additionally, if a country is associated with a matching ctrp in our library of high-risk locations, this country will now also be listed under matchingCountries.
Improved normalization of narratives: We have resolved an issue arising from the combination of Swift's 35-character line limit and the use of commas as line breakers, leading to unintended token concatenation in narrative values. This problem originated from a normalization rule intended to manage names split by commas, a common issue due to the inherent field formatting specifications of Swift MT messages. The resolution involved adjusting this rule to concatenate tokens only when a comma is detected at a position that aligns with the exact Swift field formatting specifications, thereby enhancing the accuracy of our screening process.
Thank you for choosing Screena. We look forward to bringing you more updates and innovations in future releases.