TrustSVD: Collaborative Filtering with Both the Explicit and Implicit Influence of User Trust and of Item Ratings

By Guibing Guo, Jie Zhang, Neil Yorke-Smith. 2015

598 citations recorded by the source corpus.

Collaborative filtering suffers from the problems of data sparsity and cold start, which dramatically degrade recommendation performance. To help resolve these issues, we propose TrustSVD, a trust-based matrix factorization technique. By analyzing the social trust data from four real-world data sets, we conclude that not only the explicit but also the implicit influence of both ratings and trust should be taken into consideration in a recommendation model. Hence, we build on top of a state-of-the-art recommendation algorithm SVD++ which inherently involves the explicit and implicit influence of rated items, by further incorporating both the explicit and implicit influence of trusted users on the prediction of items for an active user. To our knowledge, the work reported is the first to extend SVD++ with social trust information. Experimental results on the four data sets demonstrate that our approach TrustSVD achieves better accuracy than other ten counterparts, and can better handle the concerned issues.

Subfields: Computer Vision and Pattern Recognition, Information Systems, Transportation

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Information Systems

In Transportation

Citations in the corpus

How algorithmic confounding in recommendation systems increases homogeneity and decreases utility (2017, background): “Many algorithms use social network information to improve recommendations [13, 22, 42] but when training and evaluating these algorithms on data from social platforms, it is not clear if the models capture the true preferences of users or the biasing eects of platform features.”
Transfer Learning for Semisupervised Collaborative Recommendation (2016, background): “Successful applications include text mining [Dai et al. 2007], visual categorization [Shao et al. 2015], and social recommendation [Guo et al. 2015], etc.”
UniWalk: Explainable and Accurate Recommendation for Rating and Network Data (2017, methodology): “• TrustSVD [7] is based on matrix factorization, and the state-of-the-art method among the ones that use both a rating matrix and a social network.”
A Neural Influence Diffusion Model for Social Recommendation (2019, methodology): “’s preference and behavior are similar to her social connections [3, 23]. The social influence and social correlation among users’ interests are the foundation for building social recommender systems [7, 24, 35]. Due to the superiority of embedding based models for recommendation, most social recomm”
DiffNet++: A Neural Influence and Interest Diffusion Network for Social Recommendation (2020, methodology): “TrustSVD is such a representative model that shows state-of-the-art performance [12], [13].”
Dual Graph Attention Networks for Deep Latent Representation of Multifaceted Social Effects in Recommender Systems (2019, methodology): “Following other works [10, 16, 31, 36], we randomly choose 80% user-item interactions as training set and the remaining 20% as test set for each dataset.”
Model-Agnostic Counterfactual Reasoning for Eliminating Popularity Bias in Recommender System (2020, background): “rsonalized recommendation has revolutionized a myriad of online applications such as e-commerce [48, 50], search engines [38], and conversational systems [26, 40]. A huge number of recommender models [16, 23, 43] have been developed, for which the default optimization choice is reconstructing histor”
Self-Supervised Multi-Channel Hypergraph Convolutional Network for Social Recommendation (2021, methodology): “Upon this paradigm, a large number of social recommendation models have been developed [12, 21, 23, 55, 57, 61] and have shown stronger performance compared with general recommendation models.”
Inferring Latent Network from Cascade Data for Dynamic Social Recommendation (2016, methodology): “TrustSVD [8] incorporates trust networks with SVD++, a variant of matrix factorization modeling implicit influence from user latent factors through observed ratings.”
Graph Neural Networks in Recommender Systems: A Survey (2020, background): “Previously, some works design an additional social regularization term to constrain that the users with social relationship have similar representations [23, 38, 39, 53]; some combine the user embedding with the average representation of her friends as the overall user representation [12, 37].”