What Finance Can Learn From Dating Sites
summary
Argument: use graph databases (Neo4j) to treat relationships as first-class data so recommendation, fraud and entitlement problems become simple traversals. Demonstrations: a 10-line Cypher recommendation that matches mutual traits and location; a fraud prototype that links cards, IPs and accounts and scales to 28,000 req/s via websockets by following in-memory pointers; traversal speed claim of ~2M hops/core and 64M on 32 cores. Concrete risk: synthetic identities $250M loss avoided by linking small shared attributes. Graphs need RAM, ACID, not eventual consistency.
AI-written summary · about
chapters (48)
- 0:00Intro
- 0:23Who am I
- 1:28This talk is about graphs
- 1:42Who owns the graphs
- 2:31Graphs of love
- 5:03Meet Jeremy
- 5:22Good Friends
- 6:06Friends Graph
- 6:36Graph Databases
- 7:24Cypher
- 8:18Passion Graph
- 9:07Privacy
- 10:09Traits
- 11:04Location
- 11:31Chicago
- 12:18Location Matters
- 12:51The Graph
- 16:36Liars
- 17:00Jake
- 17:21Jonathan
- 18:17Evidence
- 19:07Spam Graph
- 19:55Professionals
- 20:18Finance
- 22:06Payment Graph
- 23:00Demographics
- 23:57Risk Analysis
- 25:25Consumer Graph
- 25:37Hierarchy
- 28:31Fake Phone Numbers
- 28:44Fraud Detection
- 30:09Speed
- 30:57Identity Management
- 32:17Access Control
- 32:37Security Graph
- 34:10Cyber Query
- 34:58Master Data
- 35:15Twitter
- 35:40Model W
- 36:29Summary
- 36:48Recommendation Engine
- 37:10Human Centipede
- 37:24Filtering Recommendations
- 38:05Movie Recommendations
- 39:28Why Graph Databases
- 40:21No Sequel Databases
- 41:50Performance
- 42:47Takeaway
related talks
47mWhat Business Can Learn from Dating
26mGraph Databases will change your (freakin') life
19mUnraveling the Web of Relationships: Understanding Graph Data
55mLet Me Graph That For You
32mUnderstanding Complexly Connected Data using Graph Databases- 19mUnraveling the Web of Relationships: Understanding Graph Data
49mModern Graph Databases
26mHow NEO4J Saved my Relationship