book structure

Advantages of recommender systems with generative models

Outline of key techniques and publications in LLM-driven RSs

Chap4-LLM Rec

4.3 Encoder-only LLM Rec

encoder-only LLMs can be used in two main architectures: as dense retrievers (c.f. Sec 4.3.1) or as cross-encoders (c.f. Sec 4.3.2).

Autoregressive LLM inputs are called prompts, which are sequences of tokens expressing a task such as top-k recommendation, rating prediction, or explanation generation.

4.5 Retrieval Augmented Rec

4.6 LLMs Representation Generation

4.7 Conversational Rec

Chap5-Multi-modal GMs for Rec

5.1 Introduction

Why?——item cold start problem, user request understanding, complex recsys scenes(virtual try-on, conversational shopping assistans)

Challenges——alignment，latent space learning

5.2 Contrastive Multimodal Rec

CLIP

Align BEfore Fuse (ALBEF)

5.3 Generative Multimodal Rec

GAN

VAE

Diffusion

5.4 Applications of Multimodal Rec

E-commerce
product visualization-virtual try on
Marketing-create personalized ad images and videos
Streaming services-Long and short-form video,music,audiobooks,podcasts,radio
Travel