Eugene Yan's blog posts detail methods for improving recommender systems beyond basic matrix factorization. The first post explores using Natural Language Processing (NLP) and graph-based approaches, drawing inspiration from word2vec and DeepWalk to learn product embeddings from graph sequences. The second post focuses on building a strong baseline recommender system using PyTorch, detailing the process of acquiring and parsing large datasets like Amazon's product reviews and extracting product relationships for use in these systems. AI
IMPACT These methods offer advanced techniques for building more effective recommender systems, moving beyond traditional approaches.
RANK_REASON The cluster discusses research papers and technical implementations for improving recommender systems using NLP and graph techniques.
- DeepWalk
- Eugene Yan
- Julian McAuley
- natural language processing
- NetworkX
- PyTorch
- recommender systems
- Tomáš Mikolov
- University of California, San Diego
- word2vec
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →