A Multifaceted Model for Cross Domain Recommendation Systems

Cross Domain Recommendation Systems using Deep Learning: A Systematic Literature Review

Recommender systems are essential engines to deliver product recommendations for e-commerce businesses. Successful adoption of recommender systems could significantly influence the growth of marketing targets. Collaborative filtering is a type of recommender system model that uses customers' activities in the past, such as ratings. Unfortunately, the number of ratings collected from customers is sparse, amounting to less than 4%. The latent factor model is a kind of collaborative filtering that involves matrix factorization to generate rating predictions. However, using only matrix factorization would result in an inaccurate recommendation. Several models include product review documents to increase the effectiveness of their rating prediction. Most of them use methods such as TF-IDF and LDA to interpret product review documents. However, traditional models such as LDA and TF-IDF face some shortcomings, in that they show a less contextual understanding of the document. This research integrated matrix factorization and novel models to interpret and understand product review documents using LSTM and word embedding. According to the experiment report, this model significantly outperformed the traditional latent factor model by more than 16% on an average and achieved 1% on an average based on RMSE evaluation metrics, compared to the previous best performance. Contextual insight of the product review document is an important aspect to improve performance in a sparse rating matrix. In the future work, generating contextual insight using bidirectional word sequential is required to increase the performance of e-commerce recommender systems with sparse data issues.

An empirical study on the effect of data sparsity and data overlap on cross domain collaborative filtering performance

Predicting the behaviour of the Earth’s ionosphere is crucial for the ground-based and spaceborne technologies that rely on it. This paper presents a novel way of inferring ionospheric electron density profiles and electron temperature profiles using machine learning. The analysis is based on the Nearest Neighbour (NNB) and Radial Basis Function (RBF) regression models. Synthetic data sets used to train and validate these two inference models are constructed using the International Reference Ionosphere (IRI 2020) model with randomly chosen years (1987–2022), months (1–12), days (1–31), latitudes (−60 to 60°), longitudes (0, 360°), and times (0–23 h), at altitudes ranging from 95 to 600 km. The NNB and RBF models use the constructed ionosonde-like profiles to infer complete ISR-like profiles. The results show that the inference of ionospheric electron density profiles is better with the NNB model than with the RBF model, while the RBF model is better at inferring the electron temperature profiles. A major and unexpected finding of this research is the ability of the two models to infer full electron temperature profiles that are not provided by ionosondes using the same truncated electron density data set used to infer electron density profiles. NNB and RBF models generally over- or underestimate the inferred electron density and electron temperature values, especially at higher altitudes, but they tend to produce good matches at lower altitudes. Additionally, maximum absolute relative errors for electron density and temperature inferences are found at higher altitudes for both NNB and RBF models.

CD-CARS: Cross-domain context-aware recommender systems

Data sparsity and data imbalance are practical and challenging issues in cross-domain recommender systems (RSs). This paper addresses those problems by leveraging the concepts which derive from representation learning, adversarial learning, and transfer learning (particularly, domain adaptation). Although various transfer learning methods have shown promising performance in this context, our proposed novel method RecSys-DAN focuses on alleviating the cross-domain and within-domain data sparsity and data imbalance and learns transferable latent representations for users, items, and their interactions. Different from the existing approaches, the proposed method transfers the latent representations from a source domain to a target domain in an adversarial way. The mapping functions in the target domain are learned by playing a min-max game with an adversarial loss, aiming to generate domain indistinguishable representations for a discriminator. Four neural architectural instances of ResSys-DAN are proposed and explored. Empirical results on real-world Amazon data show that, even without using labeled data (i.e., ratings) in the target domain, RecSys-DAN achieves competitive performance as compared to the state-of-the-art supervised methods. More importantly, RecSys-DAN is highly flexible to both unimodal and multimodal scenarios, and thus it is more robust to the cold-start recommendation which is difficult for the previous methods.

This chapter presents a brief and systematic overview of four major advanced recommender systems: group recommender systems, context-aware recommender systems, multi-criteria recommender systems, and cross-domain recommender systems. These advanced recommendations are characterized and compared in a unifying model as extensions of basic recommender systems. Future research topics and directions in the area of advanced personalized recommendations are discussed. Advanced recommender technologies will continue to advance.

Collaborative Filtering (CF) is a popular technique employed by Recommender Systems, a term used to describe intelligent methods that generate personalized recommendations. Some of the most efficient approaches to CF are based on latent factor models and nearest neighbor methods, and have received considerable attention in recent literature. Latent factor models can tackle some fundamental challenges of CF, such as data sparsity and scalability. In this work, we present an optimal scaling framework to address these problems using Categorical Principal Component Analysis (CatPCA) for the low-rank approximation of the user-item ratings matrix, followed by a neighborhood formation step. CatPCA is a versatile technique that utilizes an optimal scaling process where original data are transformed so that their overall variance is maximized. We considered both smooth and non-smooth transformations for the observed variables (items), such as numeric, (spline) ordinal, (spline) nominal and multiple nominal. The method was extended to handle missing data and incorporate differential weighting for items. Experiments were executed on three data sets of different sparsity and size, MovieLens 100k, 1M and Jester, aiming to evaluate the aforementioned options in terms of accuracy. A combined approach with a multiple nominal transformation and a "passive" missing data strategy clearly outperformed the other tested options for all three data sets. The results are comparable with those reported for single methods in the CF literature.

Evaluation Of Recommendation System On A Movie Dataset

Collaborative filtering algorithms, such as matrix factorization techniques, are recently gaining momentum due to their promising performance on recommender systems. However, most collaborative filtering algorithms suffer from data sparsity. Active learning algorithms are effective in reducing the sparsity problem for recommender systems by requesting users to give ratings to some items when they enter the systems. In this paper, a new matrix factorization model, called Enhanced SVD (ESVD) is proposed, which incorporates the classic matrix factorization algorithms with ratings completion inspired by active learning. In addition, the connection between the prediction accuracy and the density of matrix is built to further explore its potentials. We also propose the Multi-layer ESVD, which learns the model iteratively to further improve the prediction accuracy. To handle the imbalanced data sets that contain far more users than items or more items than users, the Item-wise ESVD and User-wise ESVD are presented, respectively. The proposed methods are evaluated on the famous Netflix and Movielens data sets. Experimental results validate their effectiveness in terms of both accuracy and efficiency when compared with traditional matrix factorization methods and active learning methods.

Robust machine learning is an increasingly important topic that focuses on\ndeveloping models resilient to various forms of imperfect data. Due to the\npervasiveness of recommender systems in online technologies, researchers have\ncarried out several robustness studies focusing on data sparsity and profile\ninjection attacks. Instead, we propose a more holistic view of robustness for\nrecommender systems that encompasses multiple dimensions - robustness with\nrespect to sub-populations, transformations, distributional disparity, attack,\nand data sparsity. While there are several libraries that allow users to\ncompare different recommender system models, there is no software library for\ncomprehensive robustness evaluation of recommender system models under\ndifferent scenarios. As our main contribution, we present a robustness\nevaluation toolkit, Robustness Gym for RecSys (RGRecSys --\nhttps://www.github.com/salesforce/RGRecSys), that allows us to quickly and\nuniformly evaluate the robustness of recommender system models.\n

A genetic algorithms-based hybrid recommender system of matrix factorization and neighborhood-based techniques

ABSTRACTCollaborative filtering is a widely used method in recommender systems research. However, contrary to the assumption that it relies solely on rating data, many contemporary models incorporate review information to address issues such as data sparsity. Although previous recommender systems utilised review texts to capture user preferences and item features, they often rely on a single‐embedding model to represent these features, which may limit the richness of the extracted information. Recent advancements suggest that combining multiple pre‐trained embedding models can enhance text representation by leveraging the strengths of different encoding methods. In this study, we propose a novel recommender system model, the Multi‐embedding Fusion Network for Recommendation (MFNR), which employs a multi‐embedding approach to effectively capture and represent user and item features in review texts. Specifically, the proposed model integrates Bidirectional Encoder Representations from Transformers (BERT) and its optimised variant, RoBERTa, both of which are pre‐trained transformer‐based models designed for natural language understanding. By leveraging their contextual embeddings, our model extracts enriched feature representations from review texts. Extensive experiments conducted on real‐world review datasets from Amazon.com and Goodreads.com demonstrate that MFNR significantly outperforms existing baseline models, achieving an average improvement of 9.18% in RMSE and 14.81% in MAE. These results highlight the efficacy of the multi‐embedding approach, indicating its potential for broader application in complex recommendation scenarios.

Latent space regularization for recommender systems

Recommender systems provide personalized recommendations to the users from a large number of possible options in online stores. Matrix factorization is a well-known and accurate collaborative filtering approach for recommender system, which suffers from cold-start problem for new users and items. When new users join the system, it will take some time before they enter some ratings in the system, until that time, there are not enough ratings to learn the matrix factorization model. Using auxiliary data such as user’s demographic, ratings and reviews in relevant domains, is an effective solution to reduce the new user problem. In this paper, we used the data of users activity from auxiliary domains to build domain-independent users representation that could be used to predict users ratings in the target domains. We proposed an iterative method which applied MAX-VAR generalized canonical correlation analysis (GCCA) on user’s latent factors learned from matrix factorization on each domain. Also, to improve the capability of GCCA to learn latent factors for new users, we propose a generalized canonical correlation analysis by inverse sum of selection matrices (GCCA-ISSM) approach, which provides better recommendations in cold-start scenarios. The proposed approach is extended using content-based features like topic models extracted from user’s reviews. We demonstrate the accuracy and effectiveness of the proposed approaches on cross-domain rating predictions using comprehensive experiments on Amazon and MovieLens datasets.