DPS Search Engine: Scoring, Name Matching, and Search Scenarios
ONESOURCE Denied Party Screening (DPS) is a solution from Thomson Reuters, designed for Global Trade customers worldwide. DPS allows you to screen both domestic and international trading partners - using names and/or addresses - against various sanctions and restricted party lists issued by regulatory agencies around the globe.
The search engine utilizes advanced technology and algorithms to evaluate and identify potential denied party hits. When hits surpass a specified threshold, they are flagged for further review, effectively balancing the reduction of false positives with the identification of relevant hits. Typically, most customers set this threshold at 80% to optimize accuracy and efficiency in the screening process.
Scoring
The search engine simplifies the concept of hit-scoring by providing the actual match percentage determined by the engine against a name/address match from a denied entity list. In the case where a hit is found against a region-based Sanction/Embargo list, the score does not produce a percentage, rather it specifically calls out hits that come from Sanctions (which are defined in the Sanctions/Embargos tab of DPS Settings).
Settings
The engine offers flexibility with the
Search Name Address Option (AND/OR)
and includes a feature called Search Name Weight
. This setting is relevant when using the AND
option, as it allows you to adjust how name and address components contribute to the final scoring of hits.By default, an equal weighting of 50% for both name and address means that the hit scores are averaged equally. However, you can adjust this weighting to emphasize either the name or the address. This adjustment is done using a slider on the DPS settings page, which allows you to tailor the impact of matches based on your specific requirements.
Additionally, the engine assesses the entire set of address-related fields collectively. This approach removes the need for individual percentage settings at the address and city levels, simplifying the configuration process and ensuring comprehensive evaluation of address data.
Name Matching
The engine employs various techniques to match
SearchNames
in foreign languages to English. However, it does not match any foreign language directly to another foreign language, except for matching between Chinese, Japanese, and Korean. For other languages, such as Russian, Arabic, Urdu, Thai, and others, matches are only made to English. This approach ensures effective name matching across different language systems while focusing on English as the primary point of comparison.There are 13 ways the engine matches names.
- Phonetic similarity
- Transliteration spelling differences
- Nicknames
- Missing spaces or hyphens
- Titles and honorifics
- Truncated name components
- Missing name components
- Out-of-order name components
- Initials
- Names split inconsistently across database fields
- Same name in multiple languages
- Semantically similar names
- Semantically similar names across languages