🔍 AML Entity Resolution & Matching System
AI/ML-based entity resolution for matching bank transactions to an internal AML client database. Combines semantic embeddings, fuzzy string matching, abbreviation expansion, and persistent account linking.
Internal AML Client Records
| Client ID | Client Name | Group/Parent | Risk | Aliases |
|---|---|---|---|---|
| 1001 | Priceline Group | Booking Holdings | Medium | Priceline, Booking, Booking.com |
| 1002 | JPMorgan Chase & Co | JPMorgan Chase | Low | JPMorgan, Chase Bank, JP Morgan Chase, JPM |
| 1003 | Siemens AG | Siemens | Low | Siemens Energy, Siemens Healthineers, Siemens Mobility |
| 1004 | Volkswagen Group | Volkswagen | Medium | VW, Volkswagen AG, Audi, Porsche, SEAT, Skoda |
| 1005 | Goldman Sachs Group | Goldman Sachs | High | Goldman, GS, Goldman Sachs & Co |
| 1006 | Mitsubishi UFJ Financial Group | MUFG | Medium | MUFG, Mitsubishi UFJ, MUFJ, Mitsubishi Bank |
| 1007 | Nestlé S.A. | Nestlé | Low | Nestle, Nestlé Group, Nestlé SA |
| 1008 | SoftBank Group Corp | SoftBank | High | SoftBank, Softbank, SB Group, Vision Fund |
| 1009 | Unilever PLC | Unilever | Low | Unilever NV, Unilever Group, UL |
| 1010 | Barclays PLC | Barclays | Medium | Barclays Bank, Barclays Capital, Barclays UK |
System: Semantic (MiniLM) + Jaro-Winkler + Levenshtein + Token Overlap + Core Name | Thresholds: ≥85% Accept | 55-85% Review | <55% Reject