Learning Informative Priors from Heterogeneous Domains to Improve Recommendation in Cold-Start User Domains

Abstract

In the real-world environment, users have sufficient experience in their focused domains but lack experience in other domains. Recommender systems are very helpful for recommending potentially desirable items to users in unfamiliar domains, and cross-domain collaborative filtering is therefore an important emerging research topic. However, it is inevitable that the cold-start issue will be encountered in unfamiliar domains due to the lack of feedback data. The Bayesian approach shows that priors play an important role when there are insufficient data, which implies that recommendation performance can be significantly improved in cold-start domains if informative priors can be provided. Based on this idea, we propose a Weighted Irregular Tensor Factorization (WITF) model to leverage multi-domain feedback data across all users to learn the cross-domain priors w.r.t. both users and items. The features learned from WITF serve as the informative priors on the latent factors of users and items in terms of weighted matrix factorization models. Moreover, WITF is a unified framework for dealing with both explicit feedback and implicit feedback. To prove the effectiveness of our approach, we studied three typical real-world cases in which a collection of empirical evaluations were conducted on real-world datasets to compare the performance of our model and other state-of-the-art approaches. The results show the superiority of our model over comparison models.

Potential Passenger Flow Prediction: A Novel Study for Urban Transportation Development

“…• WITF: A weighted irregular TF method which is similar as the CDTF (Hu et al 2016). For CDTF and WIFT, we leverage the passenger flow and ABS data to construct the tensor.…”

“…To connect multiple views, MVL-IV assumed that different views have distinct 'feature' matrices (i.e., {H i } m i=1 ), but correspond to the same coefficient matrix (i.e., W ). The tensor-based methods, such as (Hu et al 2013), (Hu et al 2016) (Taneja and Arora 2018) were proposed to address the cross-domain recommendation problem. They devised a cross-domain triadic factorization model to learn the triadic factors for user, item and domain, where the item dimensionality varies with domains.…”

HERS: Modeling Influential Contexts with Heterogeneous Relations for Sparse and Cold-Start Recommendation

“…Each user-item connection denotes a user's selection on an item. User-item connections can be regarded as one-class preference data (Hu et al 2016) which cannot differentiate user preferences. To handle the one-class problem, we treat the learning on the user-item interactions as a ranking problem (Rendle et al 2009).…”

On Cross-Domain Transfer in Venue Recommendation

“…Thus, CDCF is a form of transfer learning, for which the existing approaches can be divided into two categories: namely overlapping and non-overlapping CDCF. The overlapping CDCF models, such as [14,16,15,17], transfer knowledge from a source domain to a target one based on explicit links of users/items between the domains. For example, a user who has both Twitter and Foursquare accounts, overlapping CDCF aims to transfer user's preferences extracted from the user's tweets on Twitter (source domain) to improve the venue recommendation on Foursquare (target domain).…”