Wire Transactions

Data cleansing, entity resolution and enhanced due diligence

Problem & Current State

  • There are several issues with data coming from legacy systems & external systems, e.g., name & address bleeding, non-standard address formats, missing information like URL, ultimate global parent, country, state or city; incorrect address elements like zip codes or street address
  • These issues result in high false-positive rates due to missing linkages in customers’ transactions & among customers & inability to understand an entity from 360o perspective

Our Solution

  • Workflow-based & automated with analyst verification
  • Ingests data using pre-built connectors from several sources, e.g., transactions, customers data, focal banks data, customer data from external sources, watchlists
  • Supplements existing business rules with NLP & ML modules to preprocess, cleanse & harmonize originator & beneficiary names, addresses, SWIFT/BIC codes & bank names
  • ML-enabled entity resolution combines multiple representations of an entity; links internal & external data using edit distance, phonetic matching & unsupervised learning algorithms
  • Generate Single Source of Truth (SSoT) for extracted entities (name, complete address, supplemental information) using Scry’s dataset, external sources & proprietary ML modules
  • Provides user interface (UI) to review & modify system recommendations on entities and corresponding groups
  • System processes 4 million transactions per hour using parallel & distributed computing

Business Outcomes

  • Reduced False Positives by 60%
  • Increased entity resolution by 20%
  • Increased throughput of wire transactions processed by 800%
For Demo & Additional Information