An advanced entity resolution model developed during LauzHack 2024 in Switzerland that identifies and groups transactions belonging to the same external entities, even when their true identities are obscured.

This project was created in response to UBS's challenge at LauzHack 2024. Our system employs sophisticated modeling techniques to uncover hidden relationships between financial transactions, allowing for accurate entity mapping despite identity obfuscation.

• Identity Matching: Identifies when seemingly separate transactions belong to the same external entity
• Transaction Clustering: Groups related transactions to reveal patterns and relationships
• Entity Resolution: Uncovers true identities of external parties across multiple transactions
• Pattern Recognition: Detects common signatures that indicate same-entity origin

The project presented several complex technical hurdles:

1. Working with obfuscated identity data
2. Creating reliable matching algorithms despite limited identifiers
3. Balancing precision and recall in entity resolution
4. Processing and analyzing large transaction datasets efficiently

Our team tackled this ambitious challenge by:

• Analyzing transaction patterns to identify potential identity markers
• Developing algorithmic approaches to match entities across different transactions
• Creating clustering methods to group transactions by probable entity
• Testing and refining our models under tight hackathon constraints

This entity resolution system has potential applications in:

• Anti-money laundering (AML) operations
• Fraud detection
• Customer relationship management
• Regulatory compliance
• Risk assessment

• Paula Esteve
• Sergi Flores
• Clàudia Gallego

Special thanks to UBS for presenting such a challenging and educational problem, and to the LauzHack 2024 organizers for creating an outstanding event environment that fostered innovation and collaboration.

The project provided valuable experience in:

• Advanced modeling techniques
• Problem-solving under pressure
• Cross-functional team collaboration
• Handling complex financial data structures
• Entity resolution methodologies