SPR: Supervised Personalized Ranking Based on Prior Knowledge for Recommendation

  title={SPR: Supervised Personalized Ranking Based on Prior Knowledge for Recommendation},
  author={Chun Yang and Shicai Fan},
The goal of a recommendation system is to model the relevance between each user and each item through the user-item interaction history, so that maximize the positive samples score and minimize negative samples. Currently, two popular loss functions are widely used to optimize recommender systems: the pointwise and the pairwise. Although these loss functions are widely used, however, there are two problems. (1) These traditional loss functions do not fit the goals of recommendation systems… 

Figures and Tables from this paper



Learning Explicit User Interest Boundary for Recommendation

This work innovatively introduces an auxiliary score bu for each user to represent the User Interest Boundary and individually penalize samples that cross the boundary with pairwise paradigms, which achieves a hybrid loss of the pointwise and the pairwise to combine the advantages of both.

BPR: Bayesian Personalized Ranking from Implicit Feedback

This paper presents a generic optimization criterion BPR-Opt for personalized ranking that is the maximum posterior estimator derived from a Bayesian analysis of the problem and provides a generic learning algorithm for optimizing models with respect to B PR-Opt.

Improving pairwise learning for item recommendation from implicit feedback

The experiments indicate that the proposed adaptive sampler improves the state-of-the art learning algorithm largely in convergence without negative effects on prediction quality or iteration runtime.

Optimizing top-n collaborative filtering via dynamic negative item sampling

This paper proposes to dynamically choose negative training samples from the ranked list produced by the current prediction model and iteratively update the model, showing that this approach not only reduces the training time, but also leads to significant performance gains.

Deep Matrix Factorization Models for Recommender Systems

A novel matrix factorization model with neural network architecture is proposed that outperformed other state-of-the-art methods on several benchmark datasets and considers both explicit ratings and implicit feedback for a better optimization.

Neighbor Interaction Aware Graph Convolution Networks for Recommendation

A novel framework NIA-GCN is proposed, which can explicitly model the relational information between neighbor nodes and exploit the heterogeneous nature of the user-item bipartite graph, and generalize to a commercial App store recommendation scenario.

VBPR: Visual Bayesian Personalized Ranking from Implicit Feedback

This paper proposes a scalable factorization model to incorporate visual signals into predictors of people's opinions, which is applied to a selection of large, real-world datasets and makes use of visual features extracted from product images using (pre-trained) deep networks.

Self-supervised Graph Learning for Recommendation

This work explores self-supervised learning on user-item graph, so as to improve the accuracy and robustness of GCNs for recommendation, and implements it on the state-of-the-art model LightGCN, which has the ability of automatically mining hard negatives.

Unbiased Recommender Learning from Missing-Not-At-Random Implicit Feedback

It is demonstrated that the clipped estimator is expected to improve the performance of the recommender system, by considering the bias-variance trade-off and that the proposed method largely outperforms the baselines.

Efficient Neural Matrix Factorization without Sampling for Recommendation

This work derives three new optimization methods through rigorous mathematical reasoning, which can efficiently learn model parameters from the whole data with a rather low time complexity, and presents a general framework named ENMF, short for Efficient Neural Matrix Factorization.