This book focuses on the widespread use of deep neural networks and their various techniques in session-based recommender systems (SBRS). It presents the success of using deep learning techniques in many SBRS applications from different perspectives. For this purpose, the concepts and fundamentals of SBRS are fully elaborated, and different deep learning techniques focusing on the development of SBRS are studied. The book is well-modularized, and each chapter can be read in a stand-alone manner based on individual interests and needs. In the first chapter of the book, definitions and concepts related to SBRS are reviewed, and a taxonomy of different SBRS approaches is presented, where the characteristics and applications of each class are discussed separately. The second chapter starts with the basic concepts of deep learning and the characteristics of each model. Then, each deep learning model, along with its architecture and mathematical foundations, is introduced. Next, chapter 3 analyses different approaches of deep discriminative models in session-based recommender systems. In the fourth chapter, session-based recommender systems that benefit from deep generative neural networks are discussed. Subsequently, chapter 5 discusses session-based recommender systems using advanced/hybrid deep learning models. Eventually, chapter 6 reviews different learning-to-rank methods focusing on information retrieval and recommender system domains. Finally, the results of the investigations and findings from the research review conducted throughout the book are presented in a conclusive summary. This book aims at researchers who intend to use deep learning models to solve the challenges related to SBRS. The target audience includes researchers entering the field, graduate students specializing in recommender systems, web data mining, information retrieval, or machine/deep learning, and advanced industry developers working on recommender systems. Preface Aims and Scope Main Emphasis Target Audience Prerequisites Short Summary Acknowledgements Contents About the Authors Chapter 1: Introduction to Session-Based Recommender Systems 1.1 Introduction 1.2 Recommender Systems 1.3 Fundamentals of Session-Based Recommender Systems 1.3.1 Basic Concepts of SBRS 1.3.2 Challenges of SBRS 1.3.3 Session-Based vs. Sequential vs. Session-Aware Recommender Systems 1.4 Session-Based Recommender System Approaches 1.4.1 Traditional SBRS 1.4.1.1 Pattern/Rule Mining 1.4.1.2 K-Nearest Neighbors 1.4.1.3 Markov Chain 1.4.1.4 Generative Probabilistic Model 1.4.1.5 Latent Representation 1.4.2 Deep Learning SBRS 1.5 Conclusion References Chapter 2: Deep Learning Overview 2.1 Introduction 2.2 Fundamentals of Deep Learning 2.2.1 History of Deep Learning 2.2.2 AI, ML, and DL 2.2.3 Advantages of Deep Learning 2.2.4 General Process of Deep Learning-Based Solutions 2.2.5 Taxonomy of Deep Learning Models 2.3 Deep Discriminative Models 2.3.1 Multilayer Perceptron 2.3.2 Convolutional Neural Network 2.3.3 Recurrent Neural Network 2.3.3.1 LSTM 2.3.3.2 GRU 2.4 Deep Generative Models 2.4.1 Autoencoders 2.4.1.1 Sparse Autoencoder 2.4.1.2 Denoising Autoencoder 2.4.1.3 Contractive Autoencoder 2.4.1.4 Convolutional Autoencoder 2.4.1.5 Variational Autoencoder 2.4.2 Generative Adversarial Networks 2.4.3 Boltzmann Machines 2.4.3.1 Restricted Boltzmann Machine 2.4.3.2 Deep Belief Network 2.4.3.3 Deep Boltzmann Machine 2.5 Graph-Based Models 2.5.1 Graph Neural Network 2.5.2 Graph Convolutional Network 2.6 Conclusion References Chapter 3: Deep Discriminative Session-Based Recommender System 3.1 Introduction 3.2 Fundamentals 3.2.1 Datasets 3.2.2 Evaluation 3.3 Session-Based Recommender System Using RNN 3.3.1 Why RNN? 3.3.2 GRU Approaches 3.3.3 LSTM Approaches 3.4 Session-Based Recommender System Using CNN 3.4.1 Why CNN? 3.4.2 CNN Approaches 3.5 Discussion 3.6 Conclusion References Chapter 4: Deep Generative Session-Based Recommender System 4.1 Introduction 4.2 Fundamentals 4.2.1 Datasets 4.2.2 Evaluation 4.3 Session-Based Recommender System Using Autoencoder 4.3.1 Why Autoencoder? 4.3.2 Autoencoder Approaches 4.4 Session-Based Recommender System Using GAN 4.4.1 Why GAN? 4.4.2 GAN Approaches 4.5 Session-Based Recommender System Using FBM 4.5.1 Why Flow-Based Models? 4.5.2 Flow-Based Approaches 4.6 Discussion 4.7 Conclusion References Chapter 5: Hybrid/Advanced Session-Based Recommender Systems 5.1 Introduction 5.2 Fundamentals 5.2.1 Datasets 5.2.2 Evaluation 5.3 SBRS Using Hybrid Deep Neural Networks 5.3.1 Why Hybrid Deep Neural Network? 5.3.2 Approaches Based on CNN and LSTM 5.3.3 Approaches Based on CNN and GRU 5.3.4 Approaches Based on RNN and Autoencoder 5.4 SBRS Using Deep Graph Neural Network 5.4.1 Why Graph Neural Network? 5.4.2 Approaches Based on GNN 5.4.3 Approaches Based on GNN and RNN 5.4.4 Approaches Based on GCN 5.5 SBRS Using Deep Reinforcement Learning 5.5.1 Why Deep Reinforcement Learning? 5.5.2 Approaches Based on Deep Q-Learning 5.5.3 Approaches Based on DRL and RNN 5.5.4 Approaches Based on DRL and CNN 5.5.5 Approaches Based on DRL and GAN 5.6 Discussion 5.7 Conclusion References Chapter 6: Learning to Rank in Session-Based Recommender Systems 6.1 Introduction 6.2 Fundamentals 6.2.1 Ranking Creation 6.2.2 Ranking Aggregation 6.2.3 Datasets 6.3 Ranking Creation 6.3.1 Pointwise Methods 6.3.1.1 Pointwise Methods in Information Retrieval 6.3.1.2 Pointwise Methods in Recommender Systems 6.3.2 Pairwise Methods 6.3.2.1 Pairwise Methods in Information Retrieval 6.3.2.2 Pairwise Methods in Recommender Systems 6.3.3 Listwise Methods 6.3.3.1 Listwise Methods in Information Retrieval 6.3.3.2 Listwise Methods in Recommender Systems 6.3.4 Hybrid Methods 6.4 Ranking Aggregation 6.4.1 Ranking Aggregation Methods in Information Retrieval 6.4.2 Ranking Aggregation Methods in Recommender Systems 6.5 Discussion 6.6 Conclusion References Summary Index
