Building a Cross-Border KG: AI Powered Entity Resolution & Risk Detection
summary
ClearPic.ai aggregates five years of public registries, court PDFs and media across seven jurisdictions into a unified knowledge graph of ~25M data points, using phonetic matching adapted for Cyrillic and local Latin scripts, FastText/BERT embeddings and probabilistic scoring (0–1) tuned with ~500 labeled clusters. Edge-weighted graph queries run in-memory (~500 GB) enabling three‑hop traversals in sub‑seconds, raising hidden-risk detection ~40%, surfacing 15+ financial‑crime patterns and cutting checks from days to minutes with an 0.85 confidence threshold.
AI-written summary · about
chapters (7)
- 0:00Data challenges in Central Asia
- 2:25Limitations of standard tools
- 4:06Building the hybrid pipeline
- 5:21Knowledge graph fundamentals
- 13:44Data source integration
- 15:37Advanced entity resolution
- 21:45Results and future outlook
related talks
30mEntity Resolved Knowledge Graphs
32mFINRA's knowledge graphs improve regulatory investigations
29mPowering Legal Research with Knowledge Graphs and Large Language Models
19mAI-Driven Multi-Document Correlation for Financial Compliance
41mSimilar entity detection in large data
15mGraphRAG-Powered AI Agents: Transforming Corporate Metadata into Intelligent Knowledge
28mData-Empowered AI Assistants
33mSmarter Industrial Fault Diagnosis with Knowledge Graphs and Large Language Models