Doxing Dark Money: Entity Resolution to Empower AI in Anti-Fraud

Doxing Dark Money: Entity Resolution to Empower AI in Anti-Fraud

by Paco Nathan

Applied Sciences & Engineering Computing & Cybersecurity UComputing and Information TechnologyUNDatabases
🎙 Paco Nathan 👥 5K 📅 October 23, 2025 ⏱ 26 min 👁 52 📄 expert opinion 🧭 2026-08-15
Available in: English (current) Français

Keywords

entity resolutionanti-fraudgraph analyticsAIfinancial crime

Summary

Paco Nathan, Principal DevRel Engineer at Senzing, presents a talk on using entity resolution to empower AI applications in anti-fraud. He begins by highlighting the scale of financial crime, citing an estimated $3 trillion flowing through shell companies annually. He introduces the Azerbaijani laundromat case as a concrete example, where $3 billion was laundered through 500 shell companies. Nathan explains the core concept of entity resolution, which involves triangulating on PII features across multiple data sources to identify consistent footprints of individuals and organizations. He emphasizes the challenges of cultural variations in names and addresses, and the importance of evidence-based matching. He then demonstrates how entity resolution results can be used to build knowledge graphs, apply graph algorithms like centrality to identify key orchestrators, and support downstream AI applications such as summarization and tailored reporting for analysts. He also introduces open-source tools like Senzing Semantics and a sample application based on DSPI, which uses entity resolution to enhance graph RAG. The talk concludes with a demonstration of the Azerbaijani laundromat data, showing how graph algorithms can be used for forensic accounting and flow analysis.

184 words

Critical Evaluation

Value of the Information & Strength of the Argument

The talk provides valuable insights into the practical application of entity resolution in anti-fraud, supported by real-world examples and open data sources. The argumentation is solid, drawing on the speaker’s extensive experience and technical expertise. The emphasis on evidence-based matching and the importance of cultural nuances adds depth to the discussion. The presentation of open-source tools and tutorials enhances the practical value for practitioners.

Scientific Rigor, Source Quality, Title Accuracy

The speaker references several open data sources and tools, such as Open Sanctions, Open Ownership, and the Azerbaijani laundromat dataset. He also mentions his own recent publications and tutorials. However, the talk lacks formal citations to peer-reviewed literature, and the sources are primarily from the speaker’s own work or well-known investigative journalism. The title accurately reflects the content, focusing on entity resolution and its role in anti-fraud AI.

147 words

Title / Content Match

The title accurately reflects the content, focusing on entity resolution and its application in anti-fraud AI.

Quality & Reliability

8/10

The speaker is a recognized expert in the field with a strong technical background and practical experience. The talk is based on real-world case studies and open data sources, but it is primarily an expert opinion with limited peer-reviewed references.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

  • MLOps World — Conference website where the talk was presented

Concurring Sources

Contribution & Novelties

The talk provides a practical overview of entity resolution and its integration with AI for anti-fraud, highlighting open-source tools and real-world case studies. It emphasizes the importance of evidence-based identity matching and cultural awareness. The speaker introduces his own open-source libraries and tutorials, offering actionable resources for practitioners.

Pour aller plus loin :

  • Entity Resolution — Wikipedia article on record linkage, the broader concept.
  • Graph Analytics — Wikipedia article on graph analytics.
  • Open Sanctions — Open data source for sanctions and politically exposed persons.
  • Open Ownership — Open data source for beneficial ownership information.
  • DSPy — GitHub repository for DSPy, a framework for programming language models.

106 words

Radar Profile

The radar profile shows high scores in quantity and quality of information, with a slightly lower but still solid technical level. The overall reliability is high, reflecting the speaker's expertise and use of real-world examples.

Reliability 8/10

💬 No comments were provided for analysis.