Leverage side information for top‐N recommendation with latent Gaussian process

Abstract

Currently, Recommender Systems (RS) have been ubiquitously applied to various online applications and obtained tremendous success due to their capability to overcome information overload; however, the available Side Information (SI), such as demographics and attributions of items, is always neglected. Actually, SI could reflect user's interests and preference, and even influence user's decision over items; therefore, it would be greatly helpful to leverage side information to improve the performance of RS, which is already partly certificated in previous research studies. Motivated by this, in this article, a novel multi-task learning-based recommendation system referred to as LGP-RS is proposed, which takes utility functions mapping from the non-linear side information to low-rank feature space with Gaussian process (GP) priors, for users and items, respectively. In addition, Laplace approximation method is employed to approximate the posterior distribution for the utility functions, and gradient-based method is employed to learn the hyperparameters for GP. This flexible approach is able to capture the non-linear high-order interactions within side information, reduce the prediction uncertainty, and provide personalized top-N recommendation. Experimental analysis over three real-world datasets demonstrates that LGP-RS could deal with both explicit and implicit side information simultaneously, and significantly outperforms state-of-the-art approaches.

Gender In Gender Out: A Closer Look at User Attributes in Context-Aware Recommendation

“…This point may not be surprising in light of the wellknown disappointment of item attributes [18]. However, many papers studying side information combine item and user side information [9,12,33], rather than separating out user attributes as we do. Second, we show that user attributes actually have the potential to harm recommendation when we look beyond prediction performance to metrics like coverage and diversity (Section 5).…”

“…Variational Autoencoder approaches include [12], which stacks denoising auto-encoders (SDAE) to integrate side information into the latent factors, and [9], which uses a collective Variational Autoencoder (cVAE) for integrating side information for Top-N recommendation. More recent work includes a clustering-based collaborative filtering algorithm that integrates user side information (such as age, gender and occupation) in a deep neural network [32] and a Gaussian process based recommendation framework that leverages side information [33]. These approaches illustrate that researchers are interested in user attributes not just for improving cold start and sparsity, but also recommender performance across users.…”

A social hybrid recommendation system using LSTM and CNN

“…Utilizing machine learning and data mining, one can find unseen information and predict whether a user will like a particular item 7,8 . This approach has also been performed in recent research for COVID‐19 analysis and prediction 9,10 .…”

“…6 Utilizing machine learning and data mining, one can find unseen information and predict whether a user will like a particular item. 7,8 This approach has also been performed in recent research for COVID-19 analysis and prediction. 9,10 In recent years, with the rapid development of deep learning, their application in these systems, particularly with unstructured data, has enhanced.…”

No-Regret Algorithms for Private Gaussian Process Bandit Optimization

“…Gaussian Process (GP) bandit optimization (Srinivas et al, 2010) is a sequential decision problem that has a variety of human-centered applications, e.g., clinical drug trials (Costabal et al, 2019;Park et al, 2013;Peterson et al, 2017), personalized shopping recommendations (Rohde et al, 2018;Zhou et al, 2019), news feed ranking (Agarwal et al, 2018;Letham & 2019; Vanchinathan et al, 2014). It is increasingly becoming desirable that algorithms interacting with such data maintain the privacy of the individuals whose information is used (Cummings & Desai, 2018).…”