liveBook · Manning

What you will learn in AI Recommender Systems

Transformers, LLMs, agents, and RAG

by Kim Falk · Manning Early Access Program (MEAP)

Chapter by chapter

Chapter 1: Recommendations, recommender systems, and components

The introduction establishes why recommender systems matter in everyday online experiences and previews the chapter’s topics.

1.1 Starting on Recommender Systems · 1.2 What is a Recommendation · 1.2.1 User-to-item recommendations (U2I) · 1.2.2 Item-to-item recommendations (I2I) · 1.3 Building Recommenders with Different Data Types · 1.3.1 Content-based recommendations with metadata · 1.3.2 Collaborative filtering with Behavioral data · 1.3.3 Contextual data · 1.3.4 Available Datasets · 1.4 What is a Recommender System · 1.4.1 The four-stage recommender system · 1.4.2 The four-stage architecture across domains · 1.5 Further reading · 1.6 Summary

What you learn in chapter 1 →

Chapter 2: Implementing a simple recommender model

The chapter introduces simple recommender models as reference points for later evaluation and explains why personalized predictions need a user or session context.

2.1 Start with the simplest recommender system · 2.1.1 Popularity: The Fallback Recommender · 2.2 Building a Collaborative Filtering Recommender · 2.2.1 The Data: User-Item Interactions · 2.2.2 Stage 1: Retrieval - Finding Similar Items · 2.2.3 Stage 2: Filtering - Removing Consumed Items · 2.2.4 Stage 3: Scoring - Adding Multiple Signals · 2.2.5 Stage 4: Ranking - Combining Multiple Scores · 2.2.6 The four-stage recommender in action · 2.2.7 "Popular for You" - Personalized Popularity · 2.2.8 Creating the Framework · 2.2.9 The Cold Start Problem · 2.3 Content-Based Recommendations · 2.3.1 Using Item Metadata · 2.3.2 Fitting Into the Four-Stage Framework · 2.3.3 Combining Behavioral and Content Signals · 2.4 Wrapping up · 2.5 Further Reading · 2.6 Summary

What you learn in chapter 2 →

Chapter 3: Embedding-based recommender systems

Introduces embeddings as similarity representations for recommenders and contrasts behavioral, content-based, and LLM embedding sources.

3.1 Representations and embeddings · 3.1.1 Embeddings as Compressed Representations · 3.1.2 Why Embeddings Enable Similarity Search · 3.1.3 The many ways to create an Embedding · 3.2 Collaborative filtering (CF) · 3.3 Collaborative Filtering with Matrix Factorization · 3.3.1 How Matrix Factorization Works · 3.3.2 From SVD to Alternating Least Squares · 3.3.3 Implementing ALS · 3.3.4 Training an ALS Model · 3.3.5 Generating Recommendations · 3.3.6 Strategies for multiple seeds · 3.3.7 Using the framework · 3.3.8 Choosing Hyperparameters · 3.3.9 Production Considerations · 3.4 Embeddings from Content: BERT · 3.4.1 Sentence Transformers: BERT for Similarity Search · 3.4.2 Creating Content Embeddings · 3.4.3 Finding Similar Items · 3.5 Embeddings from Language Models · 3.5.1 What Makes LLM Embeddings Different? · 3.5.2 Using Open-Source LLM Embeddings · 3.5.3 Steering Embeddings with Instructions · 3.5.4 When LLM Embeddings Provide Real Value · 3.6 Combining Multiple Embedding Sources · 3.7 Summary

What you learn in chapter 3 →

Chapter 4: Evaluating what makes a good recommendation

The introduction explains why recommender evaluation must compare alternatives and help teams make product decisions.

4.1 Why Evaluation is Different for Recommender Systems · 4.1.1 There is no ground Truth · 4.1.2 Evaluation is Always Comparative · 4.1.3 What Makes a Good Recommendation? · 4.2 The Evaluation Process: From Data to Decisions · 4.2.1 The Offline Evaluation Workflow · 4.2.2 Evaluation as Experimentation · 4.2.3 What We're Actually Measuring · 4.3 Preparing Data for Evaluation · 4.3.1 Temporal Splitting: Simulating the Future · 4.3.2 Start with a sample · 4.4 Your First Evaluation: A Simple Example · 4.4.1 Setting Up the Evaluation · 4.4.2 Defining the Relevance Set · 4.4.3 Calculating Your First Metric · 4.4.4 Adding a Baseline: Popularity · 4.5 Establishing Baselines · 4.5.1 The Baseline Ladder · 4.5.2 Academic vs. Industry Baselines · 4.6 Metrics: Three Dimensions of Quality · 4.6.1 Accuracy Metrics: Precision, Recall, and Hit Rate · 4.6.2 Averages and Distributions · 4.6.3 Rank-Aware Metrics · 4.6.4 Beyond Accuracy · 4.6.5 Evaluating the Full Pipeline · 4.7 Users and Their Lifecycle · 4.8 Putting It All Together · 4.9 Further reading · 4.10 Summary

What you learn in chapter 4 →

Concepts explained in this book

These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.