Breaking Silos: Using Neo4j to Unify Tax and Finance Data
summary
Shows a working pipeline that extracts entities and ownership links from Snowflake into a Neo4j graph, runs entity resolution by comparing relationship patterns (they flag matches at 99% similarity), and surfaces anomalies across 15–16 hops without recomputing 150,000 nodes. The build includes incremental event triggers to keep the graph near-real-time, Bloom visualizations for stakeholder review, and added graph-derived features that improve downstream ML and tax-allocation calculations.
AI-written summary · about
chapters (7)
- 0:00Introduction to breaking silos
- 1:29Problem of fragmented data
- 6:03Why relationships matter in tax
- 9:52Graph architecture and entity resolution
- 17:23Data pipeline from warehouse to graph
- 23:30Breaking silos with incremental updates
- 26:39Visualizations and takeaways
related talks
27mNeo4j GDS for Snowflake
15mEasy Metadata Management (Git for Data) Using Neo4j
46mFraud Detection at Floa
08mExploring Neo4j
42mFighting fraud with graph databases
44mWhat Finance Can Learn From Dating Sites
27mDataEngBytes 2023 - SYD-T3-03 - Suzanne Nieuwenhuizen
28mNeo4j for Beginners: Pros and Cons, Based on Social Networks