Dynamic Poisson Factorization

Abstract

Models for recommender systems use latent factors to explain the preferences and behaviors of users with respect to a set of items (e.g., movies, books, academic papers). Typically, the latent factors are assumed to be static and, given these factors, the observed preferences and behaviors of users are assumed to be generated without order. These assumptions limit the explorative and predictive capabilities of such models, since users' interests and item popularity may evolve over time. To address this, we propose dPF, a dynamic matrix factorization model based on the recent Poisson factorization model for recommendations. dPF models the time evolving latent factors with a Kalman filter and the actions with Poisson distributions. We derive a scalable variational inference algorithm to infer the latent factors. Finally, we demonstrate dPF on 10 years of user click data from arXiv.org, one of the largest repository of scientific papers and a formidable source of information about the behavior of scientists. Empirically we show performance improvement over both static and, more recently proposed, dynamic recommendation models. We also provide a thorough exploration of the inferred posteriors over the latent variables.

RecSys 2015

References in corpus (1)

• Nonparametric Bayesian Factor Analysis for Dynamic Count Matrices

Cited by in corpus (19)

• Session-based Social Recommendation via Dynamic Graph Attention Networks
• Bayesian Temporal Factorization for Multidimensional Time Series Prediction
• Mixing Dirichlet Topic Models and Word Embeddings to Make lda2vec
• Deep Coevolutionary Network: Embedding User and Item Features for Recommendation
• Towards Psychologically-Grounded Dynamic Preference Models
• Link prediction in dynamic networks using random dot product graphs
• Graph link prediction in computer networks using Poisson matrix factorisation
• Recurrent Poisson Factorization for Temporal Recommendation
• A Comparative Study of Gamma Markov Chains for Temporal Non-Negative Matrix Factorization
• Deep Markov Spatio-Temporal Factorization
• Deep Poisson gamma dynamical systems
• Improving Optimization for Models With Continuous Symmetry Breaking
• Dynamic Collaborative Filtering with Compound Poisson Factorization
• Fast Non-Bayesian Poisson Factorization for Implicit-Feedback Recommendations
• Multi-level hypothesis testing for populations of heterogeneous networks
• dynamicMF: A Matrix Factorization Approach to Monitor Resource Usage in High Performance Computing Systems
• Community detection over a heterogeneous population of non-aligned networks
• A Quantum Field Theory of Representation Learning
• A Price Driven Hazard Approach to User Retention