Learning to recommend with social trust ensemble

By Hao Ma, Irwin King, Michael R. Lyu. 2009

923 citations recorded by the source corpus.

As an indispensable technique in the field of Information Filtering, Recommender System has been well studied and developed both in academia and in industry recently. However, most of current recommender systems suffer the following problems: (1) The large-scale and sparse data of the user-item matrix seriously affect the recommendation quality. As a result, most of the recommender systems cannot easily deal with users who have made very few ratings. (2) The traditional recommender systems assume that all the users are independent and identically distributed; this assumption ignores the connections among users, which is not consistent with the real world recommendations. Aiming at modeling recommender systems more accurately and realistically, we propose a novel probabilistic factor analysis framework, which naturally fuses the users' tastes and their trusted friends' favors together. In this framework, we coin the term Social Trust Ensemble to represent the formulation of the social trust restrictions on the recommender systems. The complexity analysis indicates that our approach can be applied to very large datasets since it scales linearly with the number of observations, while the experimental results show that our method performs better than the state-of-the-art approaches.

Subfields: Artificial Intelligence, Computer Networks and Communications, Information Systems

How researchers see this paper

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

In Artificial Intelligence

In Computer Networks and Communications

In Information Systems

Citations in the corpus

How algorithmic confounding in recommendation systems increases homogeneity and decreases utility (2017, background): “ularity will perform better on this data. Facebook and Etsy are among websites that use social networks to recommend content. Many algorithms use social network information to improve recommendations [13,22,26,41,42,65] but when training and evaluating these algorithms on data from social platforms,”
Comparing the Predictive Capability of Social and Interest Affinity for Recommendations (2014, background): “Although the impact of the parameters of a recommendation strategy has been often inspected [9, 39, 22, 41, 29, 20, 7], little systematic effort has been devoted into understanding how various user or item attributes are affecting the performance [2], and none of such analyses, to our knowledge, hav”
Inferring Latent Network from Cascade Data for Dynamic Social Recommendation (2016, background): “RSTE [17] predicts a user’s ratings by the linear combination of the user and their trusted friends’ latent factors vectors.”
TrustSVD: Collaborative Filtering with Both the Explicit and Implicit Influence of User Trust and of Item Ratings (2015, methodology): “…of ratings that are given by the active user and received by the target item, respectively; (2) Trust-based models: SoRec (Ma et al. 2008), RSTE (Ma, King, and Lyu 2009), SoReg (Ma et al. 2011), SocialMF (Jamali and Ester 2010), TrustMF (Yang et al. 2013) and Fang's (Fang, Bao, and Zhang 2014);…”
A survey of collaborative filtering based social recommender systems (2014, background): “Recommendation with the Social Trust Ensemble (STE) was introduced in [46].”
Exploiting geographical influence for collaborative point-of-interest recommendation (2011, background): “Recent studies have argued that social friends tend to share common interests and thus can be used in the process of collaborative filtering for making recommendations [11, 12, 2, 13, 24].”
Multi-relational matrix factorization using bayesian personalized ranking for social network data (2012, background): “STE (Social Trust Ensemble) [5] is a weighted combination of a factorization on the user-item relation and neighborhood score, derived from the sum of the dot products of the item and the friends.”
Self-Supervised Multi-Channel Hypergraph Convolutional Network for Social Recommendation (2021, background): “The common ideas of MF-based social recommendation algorithms can be categorized into three groups: co-factorization methods [22, 46], ensemble methods [20], and regularization methods [23].”
Coupled Matrix Factorization Within Non-IID Context (2014, background): “ltiplication of the two decomposed matrices. With the advent of social network, many researchers have started to analyse social recommender systems and various models integrating social networks [14] [13] [15] [28] [29] have been proposed. Social friendship is an outstanding explicit factor to impro”
Social Recommendation Using Low-Rank Semidefinite Program (2011, methodology): “Currently, a very effective approach to recommendation problem is to predict the missing data entries through factorizing the user-item rating matrix into the user-specific and item-specific matrices (Salakhutdinov and Mnih 2008; Ma et al. 2008; Ma, King, and Lyu 2009).”
The Limits of Popularity-Based Recommendations, and the Role of Social Ties (2016, background): “They are recognized to be an important factor that shape user choices, and recommenders try to leverage them in a variety of ways [22, 31, 33, 6].”
Representation learning via Dual-Autoencoder for recommendation (2017, background): “Ma et al. (2008) fused user–item matrix with users’ social trust network by sharing a common latent low-dimensional user feature matrix. Furthermore, authors in Ma, King, and Lyu (2009) coined the social trust ensemble to represent the formulation of the social trust restrictions. Some researchers h”
Music recommendation by unified hypergraph: combining social media information and music content (2010, background): “Moreover, a user’s interest may be affected by his/her friends [19, 24].”
A matrix factorization technique with trust propagation for recommendation in social networks (2010, methodology): “Similar to the STE model presented in [9], SocialMF learns Different from STE, the feature vector of each user is dependent on the feature vectors of his direct neighbors in the social network.”
Recommender systems with social regularization (2011, methodology): “RSTE : this approach is proposed in [22].”
Social Collaborative Viewpoint Regression with Explainable Recommendations (2017, background): “In social media recommendation tasks [15, 24, 50], social relations have been considered.”
DeepInf: Social Influence Prediction with Deep Learning (2018, background): “…maximization problem [24], both independent cascade and linear threshold models assume a pairwise influence model; In social recommendation [30], a key assumption is social influence—the ratings and reviews of existing users will influence future customers’ decisions through social…”
Exploring social influence for recommendation: a generative model approach (2012, methodology): “In [17, 18, 19], the authors propose to use matrix factorization method to learn the user latent vectors and item latent vectors.”
Personalized entity recommendation: a heterogeneous information network approach (2014, background): “In particular, several works leverage diﬀerent social relationships in social network such as trust relationship [18, 10], friend relationship [19, 7], and user membership [31]. to boost collaborative ﬁltering since it provides user’s like-minded neighbors.”