Deep Learning and Embedding Based Latent Factor Model for Collaborative Recommender Systems

Abstract

A collaborative recommender system based on a latent factor model has achieved significant success in the field of personalized recommender systems. However, the latent factor model suffers from sparsity problems. It is also limited in its ability to extract non-linear data features, resulting in poor recommendation performance. Inspired by the success of deep learning in different application areas, we incorporate deep learning into our proposed method to overcome the above problems. In this paper, we propose a dual deep learning and embedding-based latent factor model that considers dense user and item feature vectors. The model combines the existing deep learning and latent factor models to extract deep abstractions and non-linear feature representations of the data for rating prediction. The core idea is to map the dense user and item vectors generated by embedding techniques into dual, fully connected deep neural network architectures. In these two separate architectures, it learns the non-linear representation of the input data. The method then predicts the rating score by integrating the factors obtained from the two independent structures using the inner product. From the experimental result, we observe that the proposed model outperformed state-of-the-art existing models in real-world datasets (MovieLens ML-100K and ML-1M).

Deep Learning-Based Recommendation System: Systematic Review and Classification

“…Various deep learning algorithms have been developed to learn low-dimensional embeddings of users, items, and other features, such as DeepFM [38], [109]- [113]. DeepFM combines factorization machines with deep neural networks to model feature interactions.…”

“…A significant advancement in building effective collaborative recommender systems is the use of latent factor models [113]. These models are crucial in personalized recommendation setups.…”

A Survey of Latent Factor Models for Recommender Systems and Personalization

“…Given the complexity and rapid advancements in recommender system technologies, a comprehensive survey is necessary to gather and summarize current knowledge. While existing surveys have significantly contributed to our understanding of specific advancements, such as integrating deep learning with latent factor models [1] or exploring deeper versions of latent factor models using deep learning [2], our survey aims to provide a more holistic view of the role of latent factor models in recommender systems (see Table 1).…”

“…The algorithm thus obtained is called Stochastic Gradient Descent (SGD) 1 . Although this change compared to gradient descent seems trivial, it has profound implications, both theoretical and practical, on the behavior of the algorithm and its effects on real applications:…”

Prominent User Segments in Online Consumer Recommendation Communities: Capturing Behavioral and Linguistic Qualities with User Comment Embeddings

“…Neural network deep learning models are increasingly preferred for discovering user personality traits compared to approaches of machine learning using manually extracted features, in terms of performance, time consumption, and ability to capture hidden patterns [23]. Deep learning approaches have also been introduced to overcome the shortcomings of latent factor models in collaborative recommendation systems, which suffer from sparsity problems and a limited ability to extract non-linear features [24]. User embedding methodologies can achieve high performance in author classification tasks, accounting for a wealth of linguistic variability of individual users, yet with a significant drawback; most approaches are finding it hard to determine the degree in which embeddings capture information regarding topical, sentiment, or writing style.…”