Transfer Learning in Collaborative Filtering for Sparsity Reduction

By Weike Pan, Evan Wei Xiang, Nathan Nan Liu, Qiang Yang. 2010

409 citations recorded by the source corpus.

Data sparsity is a major problem for collaborative filtering (CF) techniques in recommender systems, especially for new users and items. We observe that, while our target data are sparse for CF systems, related and relatively dense auxiliary data may already exist in some other more mature application domains. In this paper, we address the data sparsity problem in a target domain by transferring knowledge about both users and items from auxiliary data sources. We observe that in different domains the user feedbacks are often heterogeneous such as ratings vs. clicks. Our solution is to integrate both user and item knowledge in auxiliary data sources through a principled matrix-based transfer learning framework that takes into account the data heterogeneity. In particular, we discover the principle coordinates of both users and items in the auxiliary data matrices, and transfer them to the target domain in order to reduce the effect of data sparsity. We describe our method, which is known as coordinate system transfer or CST, and demonstrate its effectiveness in alleviating the data sparsity problem in collaborative filtering. We show that our proposed method can significantly outperform several state-of-the-art solutions for this problem.

Subfields: Artificial Intelligence, 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 Information Systems

Citations in the corpus

Transfer Learning in Collaborative Filtering with Uncertain Ratings (2012, methodology): “In collaborative filtering, transfer learning methods can be adaptive (Li, Yang, and Xue 2009a; Pan et al. 2010) or collective (Singh and Gordon 2008; Li, Yang, and Xue 2009b; Yoo and Choi 2009; Cao, Liu, and Yang 2010; Pan et al.”
Transferable Contextual Bandit for Cross-Domain Recommendation (2018, methodology): “CST (Pan et al. 2010) and TCF (Pan et al. 2011), for instance, learn user latent factors and item latent factors using matrix tri-factorization.”
Recommender system application developments: A survey (2015, background): “Transfer learning techniques, which can combine the relevant data from other domains into the target domain, provide a good opportunity to handle this problem [177].”
Compressed knowledge transfer via factorization machine for heterogeneous collaborative recommendation (2015, methodology): “In order to mitigate such inconsistency, an adaptation is usually adopted in knowledge transfer methods [19,22].”
A Deep Framework for Cross-Domain and Cross-System Recommendations (2018, methodology): “In contrast, Transfer-based Approaches in CDR mainly focus on transferring latent factors [Pan et al. , 2010] or rating patterns [Gao et al. , 2013] from the source domain to the target domain.”
Speedup Matrix Completion with Side Information: Application to Multi-Label Learning (2013, background): “Although several studies exploit side information for matrix recovery [1, 2, 3, 16, 29, 32, 33], most of them focus on matrix factorization techniques, which usually result in non-convex optimization problems without guarantee of perfectly recovering the target matrix.”
Transfer Meets Hybrid: A Synthetic Approach for Cross-Domain Collaborative Filtering with Text (2019, background): “he user latent factors and hence it enables knowledge transfer. CMF has its heterogeneous [39] variants, and codebook transfer [27]. The coordinate system transfer can exploit heterogeneous feedbacks [40, 57]. Multiple source domains [32] and multi-view learning [13] are also proposed for integratin”
A General Cross-Domain Recommendation Framework via Bayesian Neural Network (2018, background): “In the first category, patterns learned from the source domain are directly transferred to the target domain to improve recommendation accuracy in the target domain [18]–[20].”
Cross-Domain Recommendation: Challenges, Progress, and Prospects (2021, background): “…et al. , 2011] Multiple contexts Multi-task learning [Lu et al. , 2018] Ratings Multi-task learning [Li et al. , 2009b] Ratings Transfer learning [Pan et al. , 2010] Heterogeneous feedback Principle coordinates [Zhao et al. , 2013] Ratings Active transfer learning [Li and Lin, 2014] Ratings…”
Cross-Domain Recommendation via Cluster-Level Latent Factor Model (2013, background): “[13] proposed the model to transform knowledge from domains which have heterogeneous forms of user feedback.”
A Multi-View Deep Learning Approach for Cross Domain User Modeling in Recommendation Systems (2015, background): “As shown in [17], the authors introduced a coordinate system transfer method for cross domain matrix factorization.”
Collaborative Filtering beyond the User-Item Matrix (2014, methodology): “[Pan and Yang 2010; Pan et al. 2010].”
Feature-based transfer learning with real-world applications (2010, background): “Web applications of transfer learning include text classification across domains [151, 47, 48, 72, 195, 207, 36, 204, 213, 203], advertising [40, 39], learn to rank across domains [10, 192, 35, 68] and multi-domain collaborative filtering [103, 102, 29, 223, 143].”
Transfer Learning with Graph Co-Regularization (2012, neutral): “2011) and collaborative filtering (Pan et al. 2010b).”
Nonparametric bayesian multitask collaborative filtering (2013, neutral): “, [18, 23, 20, 16]).”
Deep Learning Based Recommender System (2017, background): “roblems. One of the most widely studied topics in cross domain recommendation is transfer learning which aims to improve learning tasks in one domain by using knowledge transferred from other domains [40, 115]. Deep learning is well suited to transfer learning as it learn high-level abstractions tha”