Session: Machine Learning + Graph Databases for Better Recommendations
This talk will cover topics related to providing relevant recommendations to users. We don’t aim to declare one recommendation method as the best but instead highlight different approaches to enriching recommendations by combining machine learning with graph databases.
The methods we evaluate include:
- Matrix Factorization with Graph Embeddings
- Content-based TFIDF
- Cosine Similarity with AQL and User Ratings
The talk will briefly cover the methods and how we generated the distance metrics and provide notebooks that go into further detail. We will show how we integrated these findings into a frontend application for movie recommendations. The talk aims to show how pairing machine learning with graph databases can improve the quality of recommendations and offers some insights into the challenges of productionizing machine learning models.
