Recommender Systems with Generative Retrieval

This tutorial provides an overview of traditional recommender systems, their challenges, and feedback loops. It introduces the TIGER Framework, covering Transformer Index for Generative Recommenders, autoregressive model for item prediction, and semantic ID generation for items. The tutorial also delves into semantic ID generation using pre-trained text encoders and RQ-VAE, highlighting benefits such as knowledge sharing, reduction of feedback loops, and scalability. Additionally, it explore...