Machine Learning + Graph Databases for Better Recommendations
summary
A demo movie site called Arango Flix demonstrates storing similarity edges in ArangoDB and serving recommendations via AQL; content methods use TF IDF and BM25 (Arango Search), collaborative methods use Surprise SVD and sklearn, and graph approaches use PyTorch Geometric with adapters for link prediction that estimate user ratings. Notebooks, a cloud demo, and open source frontend exist, a one‑terabyte benchmark is underway, and the pipeline shows inferences syncing back to the graph for production use.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:38Speaker intro
- 6:39Graph databases overview
- 15:04Recommendation systems
- 26:05ArangoFlix demo
- 27:30Content-based filtering
- 34:55Collaborative filtering
- 38:23Graph neural networks
- 42:57ML + graph integration
- 45:02Q&A
related talks
23mGraph ML – The Next Level Of Machine Learning; Learn How To Use It And When Not To
01h 14mBuilding a Recommendation System with a Hybrid Search From Scratch
26mGraph Databases will change your (freakin') life
30mGraph Machine Learning in All Its Glory!
43mAnalytics and AI Applied to Graph Databases - Introduction and Demo
01h 21mUnlocking Personalization: A Deep Dive into Modern Recommendation Algorithms
22mAI-Powered Personalized Discovery with Graph Search and VLM-Enhanced Embeddings
38mBuilding Recommender System in PHP8