NetOwl NameMatcher, the winner of the MITRE Multicultural Name Matching Challenge, is the best-of-breed intelligent name matching software that addresses the needs of many critical applications. These applications call for accurate and fast matching of names expressed by varying transliterations and spellings. For example, name matching is used for:
- Homeland security and counter terrorism (e.g., terrorist watch lists, visa screening)
- Regulatory compliance (e.g., Know Your Customer, Politically Exposed People, OFAC)
- Anti-money Laundering (AML)
- Customer Relationship Management (CRM)
Name Matching Challenges
Name matching against a database of names, especially a large-scale, multi-ethnicity name database, poses serious challenges for many reasons including:
- multiple transliteration variants of foreign names (Abdel Fattah el-Sisi – Abdul Fatah al-Sisi)
- name order variations (Park Sol Mi – Sol Mi Park)
- nicknames (Edward – Ted – Teddy)
- initials (John Fitzgerald Kennedy – J. F. Kennedy)
- tokenization variations (Abd al-Aziz – Abdul Aziz)
- orthographic variations (Teng Hsiaoping – Teng Hsiao-P’ing)
- missing/optional name tokens (Joaquín Archivaldo Guzmán Loera – Joaquín Guzmán)
- misspellings (Barack Obama – Barak Obama)
- abbreviations (Hewlett Packard – HP)
- alternate names (Mumbai – Bombay)
- names in different languages (Hu Jintao – 胡锦涛 – 胡錦濤 – Ху Цзиньтао – هو جين تاو)
Traditional approaches to the name matching problem suffer from a precision problem (extraneous matches) as well as a recall problem (missing matches).
Revolutionary Machine Learning-based Approach
With its unique, empirically driven, machine learning-based probabilistic approach, NameMatcher provides highly accurate, scalable, and flexible automated name matching for many languages and ethnicities.
NetOwl supports name matching for many entity types:
NetOwl NameMatcher uses optimized matching models for each entity type to achieve state-of-the-art accuracy. Additionally, NameMatcher automatically detects name ethnicities that aid in achieving this level of performance.
AccurateAchieves high accuracy using intelligent, probabilistic, name matching rules derived from large-scale, real-world, multi-ethnicity name variant data.
FastPerforms name matching against tens of millions or more names with sub-second response times.
ScalableHighly scalable name matching software for Big Data environments.
MultilingualSupports multiple languages including:
- Languages in Latin-based alphabets
- Chinese (traditional and simplified)
- Persian (Farsi and Dari)
- Languages in Cyrillic alphabets
Cross-lingualSupports matching of a name in one language against names in other languages.
CustomizableAllows parameter tuning as well as addition of custom rules and dictionaries.